Universal Audio Annotation Pipeline — Test-Clip Report (SFX)

Triple-ASR ensemble · greedy decoding · expressive emotion & speaking-style captions · SFX LoRA enabled

How this pipeline works

Each audio clip is run through a multi-stage pipeline that fuses several specialist models into one structured JSON annotation:

  1. Three independent ASR systems transcribe every clip in parallel, then their outputs are reconciled by majority vote:
    • VibeVoice-ASR (microsoft/VibeVoice-ASR) — end-to-end long-form ASR with built-in diarization.
    • Parakeet TDT v3 (nvidia/parakeet-tdt-0.6b-v3) + Sortformer diarization (nvidia/diar_sortformer_4spk-v1).
    • Qwen3-ASR-1.7B + Qwen3 ForcedAligner, with speakers assigned from the Sortformer diarization.
  2. Whisper expert models analyse every utterance for raw voice attributes: emotion (laion/BUD-E-Whisper), timbre (laion/timbre-whisper) and speaking style (laion/voice-tagging-whisper).
  3. SFX LoRA sound-event detector (OpenMOSS MOSS-Audio-8B-Instruct + laion/moss-audio-sfx-lora-v4, rank-128 adapter trained on ~11k LAION soundscapes) scans each clip for overlapping medium-length sound events with timestamps and captions. These predictions are fed into MOSS as candidate sound events to verify, refine and split.
  4. MOSS-Audio-8B-Thinking listens to the audio together with all of the above context and produces the final structured annotation — reconciling the three transcripts, attributing speakers, and describing emotion, age, gender, timbre, speaking style, language and accent, plus vocal bursts and background sound events. Decoding is greedy (do_sample=False).

Expressive captions: the MOSS prompt was modified so that emotion and speaking_style are no longer single labels with a numeric intensity. Instead the model writes a short, precise caption (a few words up to ~10–12) using the EmoNet voice-emotion taxonomy, pinning intensity with modifiers like barely / slight / clearly / intensely / extreme and capturing blends of co-occurring emotions (e.g. “clearly intense anger laced with wounded disappointment”). Speaking style is captioned just as vividly (e.g. “low conspiratorial whisper”, “euphoric manic rant”, “booming drill-sergeant bark”).

SFX enabled: this report includes the SFX LoRA sound-event detection stage (laion/moss-audio-sfx-lora-v4). Its raw detections are passed into MOSS as context and are also shown per clip below. All other components run as in the original pipeline.

blade_runner_clip01

blade runner.mp4 · 0.0s–58.323s · 58.358s
timetypeannotation
0.0–58.32s sound
continuous, moderately loud sound of steady rainfall or heavy water spray, accompanied by a quiet, high-frequency cricket or insect chorus
loudness: moderate
2.56–8.4s speech
speaker speaker_0
I've seen things you people wouldn't believe.
♥ clearly somber and reflective, with a sense of melancholy and gravity
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
10.72–15.2s speech
speaker speaker_0
Attack ships on fire off the shoulder of Orion.
♥ somber, ominous, and reflective
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
17.76–23.28s speech
speaker speaker_0
I watched sea beams glitter in the dark near the ten thousand gate.
♥ somber, reflective, and slightly melancholic
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
22.08–58.32s sound
slow, mournful instrumental piece, featuring low-register cellos playing a slow, sustained, and mournful melody
loudness: moderate
28.24–32.72s speech
speaker speaker_0
All those moments will be lost in time.
♥ somber, reflective, and melancholic, with a sense of finality
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
33.76–34.4s speech
speaker speaker_0
in time.
♥ somber, reflective, and slightly melancholic
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
37.68–38.32s speech
speaker speaker_0
Like
♥ somber, reflective, and slightly melancholic
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
39.84–40.48s speech
speaker speaker_0
tears
♥ somber, reflective, and slightly melancholic
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
43.2–44.48s speech
speaker speaker_0
in rain.
♥ somber, reflective, and slightly melancholic
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
48.96–51.12s speech
speaker speaker_0
Time to die.
♥ somber, reflective, and slightly melancholic, with a sense of finality
♫ slow, deliberate, measured, theatrical, with significant pauses between phrases
adult_40s · male · en · American English · very_slow · deep, resonant, slightly gravelly, male baritone
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Noise] I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. [Noise] I watched sea beans glitter in the dark near the ten thousand gate. [Noise] All those moments will be lost in time. [Noise] Like tears. [Noise] In the rain. [Music] Time to die. [Music]
Parakeet: I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. I watched sea beams glitter in the dark near the ten hours of game. All those moments will be lost in time. Like tears in rain. Time to die.
Qwen3-ASR: I've seen things people wouldn't believe Attack ships on fire off the shoulder of Orion I watched sea beans glitter in the dark near the ten thousand gate All those moments will be lost in time like tears in rain Time to die
Raw SFX LoRA detections (6) — fed into MOSS as candidate sound events
0.0–13.3s A continuous, moderately loud sound of steady rainfall or heavy water spray creates a dense and consistent background ambiance, accompanied by a quiet, high-frequency cricket or insect chorus that adds a layer of naturalistic texture.
6.2–14.8s An older-sounding adult male with a deep, resonant, and slightly gravelly voice speaks in a slow, deliberate, and somber tone. He has a formal, almost theatrical delivery style. In the background, the continuous, moderately loud sound of steady rainfall or water spray, and a quiet, high-frequency insect chorus are clearly audible.
13.3–22.1s A second adult male voice, slightly younger-sounding with a clearer, more articulate, and mid-range quality, speaks in a matter-of-fact but slightly ominous tone, as if quoting or reciting from a text. He is positioned closer to the microphone than the first speaker. The background ambiance of moderate rainfall and faint insect sounds continues consistently throughout this segment.
21.8–35.5s A quiet, melancholic instrumental piece begins, featuring a low-register cellos playing a slow, sustained, and mournful melody. The music serves as a subtle, atmospheric underscore with a somber mood. The sounds of moderate rainfall and faint insect chirping are still audible underneath the music, along with the tail end of the previous speaker's utterance.
32.8–48.5s The first older-sounding male speaker with the deep, resonant voice and theatrical delivery style resumes speaking. His delivery is slow, measured, and imbued with a sense of gravity and finality, with significant pauses between phrases. The quiet, melancholic instrumental music, featuring low-register cellos, continues in the background, and the ambient sounds of steady rainfall and insect chirping are also consistently present.
47.3–58.0s The slow, mournful instrumental piece, led by low-register cellos, swells in volume and prominence, becoming the dominant element. The cello melody evolves into a more complex, emotionally charged passage, building in intensity. Toward the end of the segment, the voice of the older male speaker can be heard uttering a single, very quiet word just before the audio ends abruptly.
Vocal-burst candidates from the specialist locator (1, threshold 0.7) + sound-effect captioner — fed to MOSS to verify
5.78–9.14s conf 0.948 — The audio contains speech. The speaker is a male with a medium pitch. The speech is clear and understandable, but the content is not discernible. The audio contains speech, as indicated by the hint. The male speaker is likely giving a presentation or having a conversation.

blade_runner_clip02

blade runner.mp4 · 58.323s–115.069s · 56.738s
timetypeannotation
0.0–6.5s sound
A high-pitched, shimmering, synthesized chime-like sound fades out, replaced by a deep, ominous, low-frequency drone that slowly fades in, accompanied by a very quiet, intermittent crackling sound in the background.
loudness: very quiet
6.5–21.5s sound
A continuous, low-frequency, ominous drone is the primary element, layered with a higher-pitched, wavering synthesized tone creating dissonance, and faint intermittent crackling sounds persisting in the background.
loudness: quiet
21.5–36.0s sound
A dense, continuous, low-frequency ominous drone is enriched with dissonant higher-frequency overtones, and a new rhythmic, pulsing synthesized layer is introduced, resembling a distorted heartbeat or slow, wet squelching, with a subtle low-frequency rumble swelling and receding.
loudness: quiet to moderate
36.0–45.5s sound
A persistent, low-frequency orchestral drone is layered with high-pitched, shimmering synthesized chords, creating a stark and spacious contrast, with a quiet, deep, rhythmic pulse subtly discernible towards the end.
loudness: quiet
45.5–56.5s sound
High-pitched, shimmering synth pads evolve into a defined harmonic progression, layered over a low-frequency, ominous drone, creating a complex, layered harmonic texture with subtle shifts generating slow, evolving tension.
loudness: quiet to conversational
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Music]
Parakeet: —
Qwen3-ASR: The
Raw SFX LoRA detections (5) — fed into MOSS as candidate sound events
0.0–6.5s The segment begins with a high-pitched, shimmering, synthesized chime-like sound, which quickly fades. This is immediately replaced by a deep, ominous, low-frequency drone that slowly fades in, establishing a dark and tense atmosphere. A very quiet, intermittent crackling sound is audible in the background, adding a layer of subtle texture to the otherwise sparse soundscape. Universal Technical Audio Analysis: The overall loudness is very quiet, characterized by a significant swell from the drone's fade-in. The tonal balance is heavily weighted towards the sub-bass and low-mids, with the initial transient of the chime providing a brief burst of high-frequency content. The timbre of the drone is smooth and resonant, while the crackling is grainy. The spatial acoustics suggest a very wide, cavernous, and synthetic space with long reverb tails. The mix is clean, with a clear separation between the faint crackling and the emerging drone. The onset of the segment starts with an abrupt, crystalline sound before transitioning to a slow ramp-up of the drone.
6.5–21.5s A continuous, low-frequency, ominous drone is the primary element, providing a dark and steady foundation for the soundscape. Layered on top is a new, higher-pitched, and slightly wavering synthesized tone that creates a dissonant and unsettling harmony, reminiscent of a musical cue for suspense or horror. Faint, intermittent crackling sounds persist in the background, adding subtle, high-frequency texture. Universal Technical Audio Analysis: The loudness is quiet and relatively steady, with subtle dynamic shifts in the upper harmonic texture of the drone. The spectral balance is dominated by low and mid-range frequencies, creating a warm but hollow feel, with the higher drone adding tension. The drone's timbre is smooth and resonant, while the higher tone has a slightly metallic, glassy quality. The spatial acoustics are expansive and reverberant, suggesting a large, empty, or synthetic environment. There is a clear layering of frequencies, with the bass drone feeling distant and enveloping, while the higher tone feels more focused.
21.5–36.0s The soundscape is dominated by a dense, continuous, and low-frequency ominous drone. The drone's texture is enriched with dissonant, higher-frequency overtones that evolve slowly, creating a sense of foreboding and tension. Midway through the segment, a new, rhythmic, and pulsing layer is introduced, consisting of a synthesized, organic-sounding texture with a repetitive but irregular pattern, like a distorted heartbeat or slow, wet squelching. A very subtle, low-frequency rumble swells and recedes in the background. Universal Technical Audio Analysis: The loudness is quiet to moderate and builds slightly with the introduction of the new element. The tonal balance is thick in the low-end, with the pulsing layer adding intermittent mid-range complexity. The timbre of the drone is smooth but gritty with added harmonics, while the new pulse is more complex and granular. The spatial field is wide and immersive, with the pulsing sound creating a slight sense of movement without being localized to a specific point. The mix feels increasingly dense as layers are added seamlessly, with some masking occurring between the different low-frequency components.
36.0–45.5s A persistent, low-frequency orchestral drone provides a dark, atmospheric bed. Layered on top, a new musical element emerges, consisting of sustained, high-pitched, shimmering synthesized chords that float above the drone. This creates a stark and spacious contrast, building a sense of anticipation and unease. Toward the end of the segment, a quiet, deep, rhythmic pulse becomes subtly discernible, reinforcing the underlying tension. Universal Technical Audio Analysis: The overall level is quiet and relatively consistent. The spectral balance is polarized, with dominant low frequencies from the drone and delicate, airy high frequencies from the synth pads, leaving the mid-range sparser. The drone's timbre is smooth and resonant, while the synth pads are glassy and ethereal. The spatial acoustics are vast and cavernous, defined by long, diffuse reverb tails on both the drone and the pads, creating a sense of immense empty space. The separation between the low and high elements is clear.
45.5–56.5s The high-pitched, shimmering synth pads are the primary focus, now evolving into a more defined harmonic progression. The underlying low-frequency, ominous drone continues to provide a dark foundation, maintaining the suspenseful mood. The sustained pad layers create a complex, layered harmonic texture, with subtle shifts in their internal structure generating a feeling of slow, evolving tension. Universal Technical Audio Analysis: The loudness remains steady at a quiet to conversational level throughout this section. The tonal balance is split between the deep, warm bass of the drone and the bright, airy shimmer of the pads. The timbre of the pads is smooth and metallic, with a slow attack and long sustain. The spatial character is extremely wide and immersive, with heavy use of hall reverb creating a vast, synthetic acoustic space. The stereo field is filled with the pad sounds, which have a gentle, slow-moving quality. The segment ends in a fade-out, indicating a transition or the end of the piece.

love_potion_bar_scene_clip01

Love Potion No 9 Bar Scene 1.mp4 · 0.0s–10.375s · 10.423s
timetypeannotation
0.0–10.38s sound
orchestral music and audience ambiance in a reverberant space
loudness: moderate
6.08–7.6s speech
speaker speaker_0
Born to lose
♥ neutral
♫ singing, melodic and sustained
adult_40s · male · en · american_english · slow · deep, resonant, slightly gravelly
8.56–8.96s speech
speaker speaker_1
Yes
♥ neutral
♫ affirmative, concise
young_adult_20s · female · en · american_english · normal · clear, bright, slightly breathy
9.44–10.16s speech
speaker speaker_0
I've lived
♥ neutral
♫ singing, melodic and sustained
adult_40s · male · en · american_english · slow · deep, resonant, slightly gravelly
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: Born to lose. Yes. I've lived.
Parakeet: Want to lose I've lived.
Qwen3-ASR: Born to lose Yes
Raw SFX LoRA detections (2) — fed into MOSS as candidate sound events
0.0–9.5s A live musical performance unfolds in a large, reverberant space, likely a theater or concert hall. A full orchestra begins with a dramatic, swelling chord progression. Mid-segment, a male vocalist enters, singing in a clear, resonant, and powerful theatrical baritone. He is accompanied by a female vocalist who sings a brief, higher-pitched harmony with him. The overall mood is dramatic and building. A quiet, ambient room tone from the audience is audible in the background. Universal Technical Audio Analysis: The recording is moderately loud with a noticeable dynamic swell as the music and vocals enter. The spectral balance is full, with the orchestra providing a rich low-mid foundation and the vocals occupying the mid-to-high frequencies. The timbre of the orchestra is warm, while the male voice is clear and resonant. The spatial acoustics are defined by a long, diffuse reverberation characteristic of a large hall, creating a sense of space and grandeur. The mix places the vocals centrally, with the orchestra spread widely. The recording captures the live ambiance, including faint audience noise, and features significant reverb as the primary effect. The audio is clean, with good separation between the vocalists and the instrumental backing.
8.0–10.1s The male lead singer holds a long, sustained note at the end of a phrase, concluding the dramatic performance. In the background, the full orchestra plays a powerful, sustained chord, holding the harmony. The performance takes place in a large, reverberant hall, and a low-level, continuous room tone from the audience persists throughout the segment. Universal Technical Audio Analysis: The loudness is sustained at a prominent level before the audio cuts off abruptly. The spectral content is rich, with the male vocal and the orchestral chord filling the mid and upper-mid frequencies. The timbre is full-bodied and operatic for the vocal, and lush for the orchestra. The spatial character is dominated by the long, natural reverberation of the performance hall, which creates a grand, spacious feel. The recording presents a wide, centered vocal over a wide orchestral spread. Audible are the natural hall reverb and the low-level hum of the audience, characteristics of a live recording. The segment ends in a hard cut, abruptly truncating the sound.

love_potion_bar_scene_clip02

Love Potion No 9 Bar Scene 1.mp4 · 10.375s–65.923s · 55.589s
timetypeannotation
0.0–55.55s sound
Ambient restaurant chatter and clinking glasses
loudness: moderate
0.48–1.52s speech
speaker speaker_0
He trapped me by the bathroom.
♥ slightly annoyed and sarcastic
♫ conversational, slightly fast-paced, clear articulation
young_adult_20s · female · en · American English · normal · soprano, slightly soft, moderately bright, slightly breathy, slight nasal touch
6.08–13.76s speech
speaker speaker_0
Can you believe it? There's just one ugly geek after another in this place. I swear, if one more of these radio shack clerks
♥ slightly annoyed and sarcastic
♫ conversational, slightly fast-paced, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch
14.0–15.0s vocal burst
speaker speaker_0
scoff
♥ slight, barely-veiled contempt with a flicker of amusement
15.04–15.68s speech
speaker speaker_0
hit on me.
♥ slightly annoyed and sarcastic
♫ conversational, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral relaxed, mostly clear, balanced head chest, near neutral light
18.4–20.72s speech
speaker speaker_1
Can I uh uh can I buy you a drink?
♥ neutral and slightly hesitant
♫ conversational, slightly hesitant, clear articulation
adult_30s · male · en · American English · normal · baritone, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, near neutral relaxed, slight roughness
22.88–24.24s speech
speaker speaker_0
Can I ask you a question?
♥ curious and slightly playful
♫ conversational, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral relaxed, mostly clear, head mixed, light neutral
24.8–25.28s speech
speaker speaker_1
Sure.
♥ neutral and calm
♫ conversational, clear articulation
adult_30s · male · en · American English · normal · mezzo-soprano, slightly soft, near neutral relaxed, mostly clear, balanced head chest, near neutral light
25.92–32.16s speech
speaker speaker_0
Why do you think I'd be interested in you? I mean, did I look at you or or smile at you or or did I flirt with you in any way?
♥ curious and slightly playful
♫ conversational, slightly fast-paced, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, neutral slightly tense, mostly clear, balanced head chest, neutral weight
33.2–33.52s speech
speaker speaker_1
No.
♥ neutral
♫ conversational, clear articulation
adult_30s · male · en · American English · normal · mezzo-soprano, slightly soft, near neutral relaxed, mostly clear, head mixed, light neutral
33.76–42.88s speech
speaker speaker_0
Well then what makes you think I'd be interested? Are, are you somebody special? Is that it? Or do you do something really fascinating? What do you do? What do you do for a living?
♥ curious and slightly playful
♫ conversational, slightly fast-paced, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, neutral slightly tense, mostly clear, balanced head chest, neutral weight
43.68–45.52s speech
speaker speaker_1
Uh, I'm, I'm a biochemist.
♥ neutral and matter-of-fact
♫ conversational, clear articulation
adult_30s · male · en · American English · normal · baritone, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, near neutral relaxed, slight roughness
45.76–48.16s speech
speaker speaker_0
Do I look like a biochemist groupie?
♥ slightly humorous and playful
♫ conversational, clear articulation
young_adult_20s · female · en · American English · normal · mezzo-soprano, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, neutral slightly tense, mostly clear, balanced head chest, neutral weight
48.88–51.76s speech
speaker speaker_1
Well, biochemists don't have groupies.
♥ slightly annoyed and dismissive
♫ conversational, clear articulation
adult_30s · male · en · American English · normal · baritone, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, near neutral relaxed, slight roughness
52.64–54.88s speech
speaker speaker_0
And why do you think that is?
♥ curious
♫ conversational, clear articulation
young_adult_20s · female · en · American English · normal · soprano, slightly soft, moderately bright clear, slightly breathy, slight nasal touch, near neutral relaxed, mostly clear, head mixed, light neutral
54.96–55.36s speech
speaker speaker_1
it's not.
♥ neutral
♫ conversational, clear articulation
adult_30s · male · en · American English · normal · baritone, slightly soft, near neutral slightly bright, slightly breathy, slight nasal touch, near neutral relaxed, slight roughness
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: You trapped me by the bathroom. You're supposed to be... I asked him, yes, it keeps me by the bathroom and it won't let me pass. Can you believe it? It's just one ugly geek after another in this place. I swear, if one more of these Radio Shack flirts... Hey, hey, hey, hey. Can I, uh, uh, can I buy you a drink? Can I ask you a question? Sure. Why do you think I'd be interested in you? I mean, did I look at you? Or, or smile at you? Or, or did I flirt with you in any way? No. Well, then what makes you think I'd be interested? Are, are you somebody special? Is that it? Or do you do something really fascinating? What do you do? What do you do for a living? Uh, I'm, I'm a biochemist. Do I look like a biochemist groupie? Well, biochemists don't have groupies. And why do you think that is? Well, it's not.
Parakeet: He trapped me by the bathroom. Can you believe it? There's just one ugly geek after another in this place. I swear, if one more of these radio shack clerks hit on me. Can I uh uh can I buy you a drink? Can I ask you a question? Sure. Why do you think I'd be interested in you? I mean, did I look at you or or smile at you or or did I flirt with you in any way? No. Well then what makes you think I'd be interested? Are you somebody special? Is that it? Or do you do something really fascinating? What do you do? What do you do for a living? Uh I'm I'm a biochemist. Do I look like a biochemist groupie? Well, biochemists don't have groupies. And why do you think that is? Well, it's not.
Qwen3-ASR: He traps me by the bathroom He keeps me Yes keeps me by the bathroom and won't let me pass Can you believe it There's just one ugly geek after another in this place I swear if one more of these radio shack clerks Hi Hi Can I uh can I buy a drink Can I ask you a question Sure Why do you think I'd be interested in you I mean did I look at you or smile at you or did I flirt with you in any way No Well then what makes you think I'd be interested Are you somebody special Is that it Or do you do something really fascinating What do you do What do you do for a living Uh I'm I'm a biochemist Do I look like a biochemist groupie Well biochemists don't have groupies And why do you think that is Well it's not
Raw SFX LoRA detections (8) — fed into MOSS as candidate sound events
0.0–6.92s A female speaker with a high-pitched, playful, and slightly nasal voice speaks with an amused and teasing tone. Her delivery is rapid and energetic. A second female speaker with a lower, calmer voice interjects briefly in a conversational manner. Both are recorded with moderate clarity in a reverberant space, accompanied by a quiet background of indistinct chatter. Universal Technical Audio Analysis: The speech is moderately loud with natural dynamics. The voices are recorded with a clear, mid-forward character, typical of studio production, but placed within a room with noticeable reverb that creates a sense of space. The background noise is low in the mix, providing a subtle room tone. The voices are well-separated from the background. The segment starts with the speech already in progress.
7.15–10.45s A male speaker with a deep, resonant, and chest-heavy voice speaks in a deadpan, declarative tone. His delivery is slow and deliberate. His voice exhibits a noticeable and pleasant reverberation, suggesting a large, acoustically lively space. A very faint, low-level room tone is audible in the background. Universal Technical Audio Analysis: The speaker's voice is very loud and has a slow decay due to the long reverb tail, creating a significant spaciousness. The tonal balance is warm, with an emphasis on low and mid-frequencies. The recording is clear and captured at a medium distance. The reverb is the dominant effect, defining the acoustic space. The onset is clear, and the tails of his words echo in the quiet environment.
10.45–15.25s A female speaker with a high-pitched, playful voice speaks with a curious tone, asking a direct question. Her voice is followed by a brief burst of group laughter, which sounds like a small, mixed-gender group reacting with amusement. The entire scene is situated in a large, reverberant room, which adds a distinct tail to the female's voice and the laughter. Universal Technical Audio Analysis: The overall level is conversational, with a dynamic peak from the laughter. The voices are clear and forward in the mix, while the laughter has a broader, more distant quality. The recording is clean, with the room's natural reverb being the only significant effect. There is good separation between the individual voices in the laughter and the room's acoustic character.
15.25–21.61s A male speaker with a medium-pitched, clear voice speaks casually and informatively, seemingly giving instructions. His pacing is relaxed and conversational. His voice shares the same noticeable reverberant quality as the previous male speaker, suggesting they are in the same large, acoustically live environment. Very faint, indistinct background chatter provides a continuous ambient layer. Universal Technical Audio Analysis: The speech is at a loud, conversational level with natural dynamics. The voice has a balanced, clear timbre. The dominant acoustic feature is the prominent reverb tail on the voice, which colors the entire segment. The background noise is subtle and does not mask the speaker. The recording captures the room's character effectively without being noisy.
22.1–28.51s A sequence of short, clipped exchanges occurs. A first male speaker (medium-pitched) asks a question. He is followed by a male with a very deep, resonant voice who answers with a short, declarative response. After a brief pause, the first male speaker offers another, slightly longer response. All are recorded in the same large, reverberant hall. Faint room tone is present. Universal Technical Audio Analysis: The loudness is consistent at a conversational level. The voices have distinct spectral profiles: the first is mid-range focused, the second is boomy and rich in low frequencies. The reverb is a constant and defining element of the recording, creating a spacious feel. The editing consists of clean cuts between the rapid succession of speakers.
29.81–36.11s A female speaker with a clear, medium-high pitched voice speaks with a direct, inquisitive tone. Her pacing is steady and her articulation is precise. She is recorded in a large, reverberant space with a male speaker's very deep, resonant voice providing a brief response toward the end of the segment. A quiet layer of indistinct room tone is audible throughout. Universal Technical Audio Analysis: The female voice is the dominant element, mixed at a loud, clear level. The tonal balance is bright and present. The reverb is consistently applied, defining the acoustic environment. The male response is quieter and lower in frequency but shares the same spatial characteristics. The recording is clean with no distracting artifacts.
36.11–44.41s A rapid-fire exchange of questions and answers takes place between two speakers in a large, reverberant room. The first speaker is a female with a medium-high pitched, inquisitive voice. The second speaker is a male with a markedly deep, resonant, and slightly boomy voice who responds in a calm, declarative manner. Faint, indistinct room tone is present in the background. Universal Technical Audio Analysis: This segment is dynamically active with loud, clear speech. The female voice is bright and forward, contrasting with the male voice's warm, bass-heavy timbre. Both voices are saturated with the same long reverb tail, glued together by the acoustic space. The editing is seamless, capturing the natural flow of the quick dialogue.
49.36–55.22s A short, quiet conversational exchange occurs between three speakers in sequence within a large, reverberant space. First, a male with a medium-pitched voice speaks. Then a female with a medium-high pitched voice responds. Finally, the first male speaker makes a brief comment. A quiet, steady room tone is audible underneath. Universal Technical Audio Analysis: The overall level is conversational but quieter than the previous segment. The tonal characteristics of the three distinct voices are clearly audible. The dominant acoustic feature is the consistent, spacious reverb that colors all the speech. The recording is clear with a low noise floor, highlighting the room's natural ambience and the subtle nuances of the dialogue.

love_potion_bar_scene_clip03

Love Potion No 9 Bar Scene 1.mp4 · 65.923s–122.887s · 56.999s
timetypeannotation
0.0–2.0s speech
speaker speaker_0
Not a very glamorous job.
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
0.0–56.96s sound
restaurant ambience with continuous chatter and clinking sounds
loudness: moderate
2.0–2.0s speech
speaker speaker_1
No.
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
3.0–3.0s speech
speaker speaker_1
It's
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
3.0–6.0s speech
speaker speaker_1
not one of your drop-dead glamorous jobs.
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
6.0–6.0s speech
speaker speaker_0
I
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
6.0–7.0s speech
speaker speaker_1
mean,
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
7.0–7.0s speech
speaker speaker_0
I'm
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
7.0–10.0s speech
speaker speaker_1
sure that you find chemicals very fascinating.
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
10.0–10.0s speech
speaker speaker_0
But
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
10.0–17.0s speech
speaker speaker_1
do I really look like somebody who'd be interested in a uh uh a chemical person?
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
19.0–19.0s speech
speaker speaker_0
No,
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
19.0–24.0s speech
speaker speaker_1
wait. I'm only doing this for your benefit. I mean,
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
24.0–24.0s speech
speaker speaker_0
you
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
24.0–26.0s speech
speaker speaker_1
wouldn't want this to happen again, would you?
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
26.0–27.0s speech
speaker speaker_0
No.
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
27.0–35.0s speech
speaker speaker_1
No. I mean, see, what is very obvious to me is not to you. I look at your clothes and then look at mine. Do you see any kind of a difference here?
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
37.0–47.0s speech
speaker speaker_1
Clothes are not important to you, but they're important to me. Taste is very important to me. I like designer clothes. I like nice things. I drive a BMW. Oh, what about you?
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
47.0–49.0s speech
speaker speaker_0
Yeah, I drive a Volkswagen.
♥ neutral
♫ casual
adult_30s · male · en · American · normal · resonant
51.0–52.0s speech
speaker speaker_1
Farfignougan.
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
52.0–52.0s vocal burst
speaker speaker_1
laugh
♥ lighthearted
53.0–56.0s speech
speaker speaker_1
You know what that really means? Can't afford the Mercedes.
♥ amused
♫ flirtatious
adult_30s · female · en · American · normal · slightly breathy
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: Not a very glamorous job. No, it's not one of your drop dead glamorous jobs. I mean, I'm sure that you find chemicals very fascinating. But do I really look like somebody who'd be interested in a, uh, a, a chemical person? No, wait, wait. I'm only doing this for your benefit. I mean, you wouldn't want this to happen again, would you? No. No. I mean, see, what is very obvious to me is not to you. I, look at your clothes, and then look at mine. Do you see any kind of a difference here? Clothes are not important to you, but they're important to me. Taste is very important to me. I like designer clothes. I like nice things. I drive a BMW. Oh, what about you? Yeah, I drive a Volkswagen. Barf McNugget. You know what that really means? Can't afford the Mercedes.
Parakeet: Not a very glamorous job. No. It's not one of your drop-dead glamorous jobs. I mean, I'm sure that you find chemicals very fascinating. But do I really look like somebody who'd be interested in a uh uh a chemical person? No, hey, wait. I'm only doing this for your benefit. I mean, you wouldn't want this to happen again, would you? No. No. See, what is very obvious to me is not to you. I look at your clothes and then look at mine. Do you see any kind of a difference here? Clothes are not important to you, but they're important to me. Taste is very important to me. I like designer clothes. I like nice things. I drive the BMW. Uh what about you? Uh drive a Volkswagen. Far Fignogan. You know what that really means? Can't afford the Mercedes.
Qwen3-ASR: Not a very glamorous job No it's not one of your drop dead glamorous jobs I mean I'm sure that you find chemicals very fascinating but do I really look like somebody who'd be interested in a a chemical person No wait I'm only doing this for your benefit I mean you wouldn't want this to happen again would you No No See what is very obvious to me is not to you Look at your clothes and then look at mine Do you see any kind of a difference here Clothes are not important to you but they're important to me Taste is very important to me I like designer clothes I like nice things I drive a BMW What about you I drive a Volkswagen Farfignougan You know what that really means Can't afford the Mercedes
Raw SFX LoRA detections (4) — fed into MOSS as candidate sound events
0.0–15.5s A conversation between two speakers in a restaurant. It begins with a male speaker with a clear, resonant adult voice speaking at a medium distance. After a pause, a female speaker responds with a higher-pitched, slightly breathy, and theatrical adult voice, speaking at a closer proximity with an amused and flirtatious tone. In the background, a continuous, low-level murmur of indistinct chatter and clinking cutlery suggests a busy dining environment. Universal Technical Audio Analysis: The overall loudness is conversational and relatively steady. The spectral balance is mid-forward, capturing the voices clearly against the ambient noise. The timbre of the male voice is smooth, while the female's is slightly airy. The spatial acoustics suggest a live, moderately reverberant indoor space, likely a dining room. The recording has a natural feel, with clear foreground-background separation and no obviously applied effects.
15.5–30.5s A conversation between two speakers in a restaurant setting. A female speaker with a clear, slightly breathy adult voice speaks for the majority of the segment, her tone becoming more pointed and defensive as she continues her thought. Her delivery is expressive and varied. Toward the end of the segment, a male speaker with a resonant adult voice interjects with a brief, firm, and dismissive "No." The persistent background consists of a diffuse, unintelligible walla of crowd murmur and occasional light clinking sounds. Universal Technical Audio Analysis: The loudness remains at a consistent conversational level with natural dynamic peaks in the female speaker's voice. The tonal balance is centered on the midrange frequencies of the speech. The spatial characteristics are defined by a noticeable room reverb that places the speakers in a shared, lively space. There is clear separation between the foreground dialogue and the background ambience. No editing or effects are apparent.
30.5–43.5s A dialogue between two speakers in a restaurant. A male speaker with a resonant adult voice speaks in a low, challenging tone, with his delivery slowing down for dramatic effect. Following his statement, a female speaker with a clear, slightly breathy adult voice responds with a taunting and assertive delivery, her pitch rising and falling expressively. The ambient sound of a busy restaurant, characterized by a constant, low-level murmur of indistinct chatter, is audible throughout. Universal Technical Audio Analysis: The loudness of the dialogue is at a conversational level. The male voice is fuller, while the female voice has more high-frequency presence, creating a distinct spectral separation. The timbre of the male voice is smooth; the female's is sharper. The acoustic space is clearly audible through the diffuse reverberation of the background noise. The recording feels untreated and captures the natural room sound effectively.
43.5–57.5s A conversational exchange between two speakers in a restaurant. A male speaker with a resonant adult voice speaks in a casual, slightly mumbled manner. A female speaker with a clear, slightly breathy adult voice immediately responds in a light-hearted, amused, and teasing tone, punctuating her speech with a brief laugh. The continuous, low-level, and unintelligible murmur of the restaurant's background ambience persists underneath the dialogue. Universal Technical Audio Analysis: The loudness is conversational, with a dynamic peak from the female's laugh. The spectral content is focused in the vocal midrange. The spatial acoustics are those of a live, reverberant room, with the voices feeling present and close despite the ambient background. The timbres of the two voices are distinct, allowing for easy separation. No significant mixing or mastering features are detected.

love_potion_bar_scene_clip04

Love Potion No 9 Bar Scene 1.mp4 · 122.887s–162.821s · 39.941s
timetypeannotation
0.0–39.91s sound
Continuous background ambiance of a large, reverberant space with a low-level murmur of indistinct crowd chatter and a faint, shimmering synthesized pad drone.
loudness: moderate
2.31–10.87s speech
speaker speaker_0
Well, you're right. We're not really suited for each other, and, uh, so I'll just go and get myself a chemical girl.
♥ mild resignation and a hint of awkwardness
♫ casual, slightly hesitant, conversational
adult_30s · male · en · American English · normal · baritone, slightly soft and neutral
11.27–12.67s speech
speaker speaker_1
I'm sorry.
♥ sincere apology, slightly distressed
♫ apologetic, soft, and clear
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
13.27–15.87s speech
speaker speaker_2
I was just trying to make a point. I tell you what,
♥ slightly annoyed, trying to explain
♫ casual, slightly insistent, conversational
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
13.27–13.87s vocal burst
speaker speaker_2
sigh
♥ mild exasperation and a hint of frustration
16.57–20.17s speech
speaker speaker_2
I'll go home with you tonight and be your total sex slave.
♥ playful, slightly exaggerated, and assertive
♫ casual, slightly dramatic, conversational
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
21.77–24.97s speech
speaker speaker_2
If you can answer one question correctly.
♥ neutral, slightly challenging, and direct
♫ casual, direct, and clear
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
27.19–27.59s speech
speaker speaker_0
All right.
♥ mild resignation and acceptance
♫ casual, slightly resigned, conversational
adult_30s · male · en · American English · normal · baritone, slightly soft and neutral
28.44–29.84s speech
speaker speaker_2
Who designed this top?
♥ direct, slightly challenging, and curious
♫ casual, direct, and clear
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
33.56–35.06s speech
speaker speaker_0
I wouldn't know.
♥ mild confusion and a hint of awkwardness
♫ casual, slightly hesitant, conversational
adult_30s · male · en · American English · normal · baritone, slightly soft and neutral
36.33–37.63s speech
speaker speaker_2
Bye-bye now.
♥ playful, slightly dismissive, and conclusive
♫ casual, slightly playful, and clear
young_adult_20s · female · en · American English · normal · soprano, slightly soft and neutral
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Human Sounds] Well, you're right. We're not really suited for each other, and, uh, so I'll just go and get myself a chemical girl. I'm sorry. I was just trying to make a point. I tell you what, I'll go home with you tonight and be your total sex slave if you can answer one question correctly. All right. Who designed this top? [Human Sounds] I wouldn't know. Bye-bye now. [Environmental Sounds]
Parakeet: Well, you're right. Uh we're not really suited for each other, and uh so I'll just go and get myself a chemical girl. I'm sorry. I was just trying to make a point. I tell you what, I'll go home with you tonight and be your total sex slave. If you can answer one question correctly.
Qwen3-ASR: Well you're right We're not really suited for each other and so I'll just go and get myself a chemical girl I'm sorry I was just trying to make a point I tell you what I'll go home with you tonight and be your total sex slave If you can answer one question correctly All right Who designed this top I wouldn't know But by now
Raw SFX LoRA detections (4) — fed into MOSS as candidate sound events
0.0–7.5s A quiet, breathy laugh from a female-sounding individual is heard, followed by a male speaker with a clear, medium-pitched, and relaxed voice. He speaks at a steady, conversational pace. In the background, a soft, shimmering synthesized pad drone provides a gentle musical underscore. Faint, indistinct chatter from a distant crowd is audible, creating a sense of a large, open, and somewhat busy public space. Universal Technical Audio Analysis: The overall level is conversational and remains steady. The tonal balance is centered in the mid-range, with the speaker's voice being the most prominent element. The timbre of the voice is natural and clear, while the background pad is smooth and airy. The spatial acoustics suggest a large, reverberant hall with a long decay time, giving the recording a sense of spaciousness. There is clear separation between the foreground speech and the more distant background layers. The onset of the speech is clean, and the segment ends as the speaker finishes his phrase.
7.0–17.5s A female speaker with a breathy, close-miked, and playful voice begins speaking. A male speaker with a deeper, more resonant voice overlaps with her, his tone shifting from playful to mock-offended and then assertive. Their dialogue exhibits a light, teasing dynamic. A quiet, shimmering musical pad and the constant, low-level murmur of distant, indistinct crowd chatter continue in the background throughout their exchange. Universal Technical Audio Analysis: The loudness is at a conversational level but increases slightly in intensity during the male speaker's more forceful lines. The spectral balance is mid-forward due to the close-miked voices. The female voice has a soft, airy timbre with noticeable breathiness, while the male voice is fuller in the low-mids. The spatial acoustics remain consistent with a large, reverberant hall, though the proximity of the voices makes them sound relatively dry compared to the environment. The dialogue is clearly layered over the constant ambient bed. The segment features a clear conversational turn
17.0–27.5s A female speaker with a clear, direct, and medium-pitched voice speaks with a steady, unwavering cadence, delivering what sounds like a challenge or condition. Her voice is prominent in the mix. A male speaker's voice is heard in the background uttering a single, quiet word of agreement. A continuous, low-level ambient layer of indistinct crowd murmur persists throughout the segment. Universal Technical Audio Analysis: The overall level is conversational and very steady, with a direct and controlled dynamic character. The tonal balance is clear and focused in the mid-range, emphasizing the female speaker's crisp articulation. The timbre of her voice is firm and articulate, contrasting with the soft, distant quality of the background elements. The spatial acoustics feature distinct layering, with the primary speaker sounding relatively dry and upfront, while the background ambience provides a sense of a large, reverberant space. The speech is cleanly separated from the background.
27.0–38.8s A female speaker with a direct, medium-pitched voice poses a question, and a male speaker responds with a hesitant utterance. The female speaker then delivers a calm, conclusive statement that acts as a finality. Her speech is clear and well-defined. A brief, metallic rattling sound occurs mid-segment after the male speaker's response. Throughout, a low-level, continuous ambience of indistinct crowd murmur is present. Universal Technical Audio Analysis: The loudness is at a consistent conversational level. The spectral balance is focused on the vocal mid-range, ensuring clarity. The timbre of the voices is natural and unadorned. The spatial acoustics are defined by the large, reverberant character of the background ambience, which contrasts with the drier quality of the foreground speech. The segment ends with an abrupt cut of all audio following the final spoken word, creating a hard and sudden offset. The speech layers are clearly separated from the constant background noise.

love_potion_monkey_test_clip01

Love Potion No. 9 - Monkey Test.mp4 · 0.0s–43.578s · 43.624s
timetypeannotation
0.0–2.64s sound timeline fill (SFX fallback)
Her pitch is relatively high, and her pace is moderate. A very quiet, steady room tone is audible underneath her speech. The tonal balance is mid-forward, emphasizing the vocal frequencies. The vocal timbre is smooth with a slight airy quality. The acoustic space is a small, untreated room, evident from the short reverb tails. There is a faint, low-level hiss present. The recording captures natural sibilance without harshness
loudness: moderate
2.64–6.32s speech
speaker speaker_0
Are you are you sure? Are you absolutely positive?
♥ clearly questioning and slightly hesitant
♫ insistent questioning, slightly halting
adult_30s · female · en · American English · normal · mezzo-soprano, clear, slightly breathy
6.32–8.88s speech
speaker speaker_1
I'm positive. Nothing to worry about. It's a it's a salt.
♥ calm and reassuring, slightly hesitant
♫ reassuring statement, slightly hesitant
adult_30s · male · en · American English · normal · baritone, warm, slightly gravelly
9.28–11.36s speech
speaker speaker_1
It's a condensed liquid salt.
♥ calm and informative
♫ informative statement
adult_30s · male · en · American English · normal · baritone, warm, clear
11.36–12.8s sound timeline fill (SFX fallback)
A faint, steady room tone provides the background ambiance. The vocal timbre is resonant with a subtle rasp. The acoustic signature is that of a small, slightly live room with short decay times. A quiet noise floor is present. The recording is clean, with minimal artifacts
loudness: moderate
12.8–14.0s speech
speaker speaker_1
You want me to take it?
♥ neutral, questioning
♫ questioning
adult_30s · male · en · American English · normal · baritone, clear
14.0–15.2s speech
speaker speaker_0
No.
♥ definitive and slightly dismissive
♫ short, definitive response
adult_30s · female · en · American English · normal · mezzo-soprano, clear
15.2–16.5s sound timeline fill (SFX fallback)
near silence with faint room tone
loudness: quiet
16.5–19.28s sound timeline fill (SFX fallback)
A distinct scraping sound if a stiff-bristled brush or broom is being pulled across a rough, hard surface like concrete or a floor. The sound is textured and rhythmic, with variations in pressure and speed. It is the primary focus of the segment. In the background, a very quiet, steady room tone is audible. Universal Technical Audio Analysis: The loudness of the scraping is moderately loud, with dynamic variations corresponding to the action. The sound is spectrally complex, with significant energy in the mid and high frequencies, creating a harsh, abrasive texture. The timbre is dry and gritty. Spatially, the sound is centered and close to the microphone. No motion is perceived. The onset is abrupt, and it maintains a relatively consistent presence until it fades out
loudness: moderate
19.28–19.6s speech
speaker speaker_0
Okay.
♥ neutral, slightly resigned
♫ neutral response
adult_30s · female · en · American English · normal · mezzo-soprano, clear
19.6–25.0s sound timeline fill (SFX fallback)
A distinct scraping sound if a stiff-bristled brush or broom is being pulled across a rough, hard surface like concrete or a floor. The sound is textured and rhythmic, with variations in pressure and speed. It is the primary focus of the segment. In the background, a very quiet, steady room tone is audible. Universal Technical Audio Analysis: The loudness of the scraping is moderately loud, with dynamic variations corresponding to the action. The sound is spectrally complex, with significant energy in the mid and high frequencies, creating a harsh, abrasive texture. The timbre is dry and gritty. Spatially, the sound is centered and close to the microphone. No motion is perceived. The onset is abrupt, and it maintains a relatively consistent presence until it fades out
loudness: moderate
20.05–20.05s vocal burst
speaker speaker_0
chuckle
♥ slightly amused, nervous
25.0–30.0s sound timeline fill (SFX fallback)
near silence with faint room tone
loudness: quiet
30.0–38.0s sound timeline fill (SFX fallback)
A sharp, forceful inhalation, sounding like a sniff or a deep breath of relief, occurs at the beginning of the segment. This is immediately followed by a series of quiet, subtle, intermittent rustling and handling noises, suggesting gentle movement or shifting of lightweight objects. A quiet, steady room tone is the underlying ambient layer. Universal Technical Audio Analysis: The overall loudness is quiet, punctuated by the moderately loud transient of the inhalation. The spectral content is varied, with the inhalation being low-mid focused and the rustles containing higher-frequency textures. The timbre of the inhalation is slightly nasal the rustles are soft and fabric-like. The acoustic space sounds like a small, quiet room. The segment begins with an abrupt, sharp transient
loudness: moderate
38.0–42.38s sound timeline fill (SFX fallback)
A loud, artificial telephone ringtone plays. It has a sequence of short, percussive, metallic-toned synthesized notes in a descending pattern. The sound is abrupt and electronic. The underlying quiet room tone is briefly exposed. Universal Technical Audio Analysis: The segment has a wide dynamic range, starting with the moderately loud ringtone and ending with the quiet speech. The ringtone is spectrally focused in the mid-high frequencies, giving it a bright, piercing quality. The vocal element is spectrally darker and lower in volume. The timbre of the ring is synthetic and metallic. Spatially, both sounds are centered and close-miked. The ringtone has a very sharp, electronic onset and offset
loudness: moderate
42.38–43.58s speech
speaker speaker_0
Didn't sound unusual.
♥ neutral, slightly dismissive
♫ neutral, slightly dismissive statement
adult_30s · female · en · American English · normal · mezzo-soprano, clear
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Silence] Are you, are you sure? Are you absolutely positive? I'm positive. Nothing to worry about. It's a, it's a salt. It's a condensed liquid salt. You want me to take it? No. [Silence] Okay. [Human Sounds] Can you turn this off?
Parakeet: Are you are you sure? Are you absolutely positive? I'm positive. Nothing to worry about. It's a it's a salt. It's a condensed liquid salt. You want me to take it? No. Okay.
Qwen3-ASR: Are you Are you sure Are you absolutely positive I'm positive Nothing to worry about It's a It's salt It's condensed liquid salt You want me to take it No Didn't sound unusual
Raw SFX LoRA detections (5) — fed into MOSS as candidate sound events
0.0–6.0s A young adult female with a clear, slightly breathy voice speaks in an insistent and questioning tone. Her pitch is relatively high, and her pace is moderate. Her voice is recorded with noticeable proximity, and the quiet room she's in has a short, bright reverberation. A very quiet, steady room tone is audible underneath her speech. Universal Technical Audio Analysis: The recording is at a conversational loudness with a clear signal. The tonal balance is mid-forward, emphasizing the vocal frequencies. The vocal timbre is smooth with a slight airy quality. The acoustic space is a small, untreated room, evident from the short reverb tails. There is a faint, low-level hiss present. The recording captures natural sibilance without harshness.
6.0–15.5s An adult male with a low-pitched, slightly gravelly voice speaks in a calm and reassuring, yet somewhat hesitant, manner. His delivery is steady and at a conversational pace. The recording captures his voice with clear presence, situated in a quiet indoor space with audible room reflections. A faint, steady room tone provides the background ambiance. Universal Technical Audio Analysis: The loudness is at a conversational level. The spectral balance is warm, with a focus on the fundamental frequencies of his male voice. The vocal timbre is resonant with a subtle rasp. The acoustic signature is that of a small, slightly live room with short decay times. A quiet noise floor is present. The recording is clean, with minimal artifacts.
16.5–25.0s A distinct scraping sound, as if a stiff-bristled brush or broom is being pulled across a rough, hard surface like concrete or a floor. The sound is textured and rhythmic, with variations in pressure and speed. It is the primary focus of the segment. In the background, a very quiet, steady room tone is audible. Universal Technical Audio Analysis: The loudness of the scraping is moderately loud, with dynamic variations corresponding to the action. The sound is spectrally complex, with significant energy in the mid and high frequencies, creating a harsh, abrasive texture. The timbre is dry and gritty. Spatially, the sound is centered and close to the microphone. No motion is perceived. The onset is abrupt, and it maintains a relatively consistent presence until it fades out.
30.0–38.0s A sharp, forceful inhalation, sounding like a sniff or a deep breath of relief, occurs at the beginning of the segment. This is immediately followed by a series of quiet, subtle, intermittent rustling and handling noises, suggesting gentle movement or shifting of lightweight objects. A quiet, steady room tone is the underlying ambient layer. Universal Technical Audio Analysis: The overall loudness is quiet, punctuated by the moderately loud transient of the inhalation. The spectral content is varied, with the inhalation being low-mid focused and the rustles containing higher-frequency textures. The timbre of the inhalation is slightly nasal, while the rustles are soft and fabric-like. The acoustic space sounds like a small, quiet room. The segment begins with an abrupt, sharp transient.
38.0–42.8s A loud, artificial telephone ringtone plays. It has a sequence of short, percussive, metallic-toned synthesized notes in a descending pattern. The sound is abrupt and electronic. After the ringtone stops, a young adult male voice speaks a single, quiet word in a neutral tone. The underlying quiet room tone is briefly exposed. Universal Technical Audio Analysis: The segment has a wide dynamic range, starting with the moderately loud ringtone and ending with the quiet speech. The ringtone is spectrally focused in the mid-high frequencies, giving it a bright, piercing quality. The vocal element is spectrally darker and lower in volume. The timbre of the ring is synthetic and metallic. Spatially, both sounds are centered and close-miked. The ringtone has a very sharp, electronic onset and offset.

love_potion_monkey_test_clip02

Love Potion No. 9 - Monkey Test.mp4 · 43.578s–57.775s · 14.237s
timetypeannotation
0.0–2.0s sound timeline fill (SFX fallback)
Her delivery is slow and deliberate, with a sense of vulnerability. The recording is captured at a medium proximity in what sounds like a small, untreated room. Throughout the segment, a quiet but noticeable background hiss and room tone are audible. The spectral balance is focused on the mid-range, with a touch of low-mid warmth, lacking significant high-frequency air. The spatial acoustics suggest a small, slightly live room with short, natural reverberation. The recording sounds unprocessed, with the prominent room tone and hiss indicating a raw or ambient capture rather than a clean studio recording
loudness: moderate
2.0–4.64s speech
speaker speaker_0
How did it sound to you?
♥ neutral, slightly questioning
♫ conversational, moderate pace
adult_30s · female · en · American English · normal · mezzo-soprano, clear, slightly soft
4.72–6.08s speech
speaker speaker_1
Well, I wouldn't know.
♥ hesitant, slightly dismissive
♫ conversational, slightly hesitant
adult_30s · male · en · American English · normal · baritone, resonant, slightly rough
4.72–6.08s vocal burst
speaker speaker_1
scoff
♥ slight contempt, dismissive
6.08–7.68s sound timeline fill (SFX fallback)
Her delivery is slow and deliberate, with a sense of vulnerability. The recording is captured at a medium proximity in what sounds like a small, untreated room. Throughout the segment, a quiet but noticeable background hiss and room tone are audible. The spectral balance is focused on the mid-range, with a touch of low-mid warmth, lacking significant high-frequency air. The spatial acoustics suggest a small, slightly live room with short, natural reverberation. The recording sounds unprocessed, with the prominent room tone and hiss indicating a raw or ambient capture rather than a clean studio recording
loudness: moderate
7.68–8.08s speech
speaker speaker_0
Is that
♥ neutral, questioning
♫ conversational, brief utterance
adult_30s · female · en · American English · normal · mezzo-soprano, clear
8.08–10.0s speech
speaker speaker_1
Is that an unusual sound for a chimp?
♥ questioning, slightly curious
♫ conversational, questioning intonation
adult_30s · male · en · American English · normal · baritone, resonant
10.0–10.96s sound timeline fill (SFX fallback)
The same quiet background hiss and room tone are present beneath the speech. The vocal timbre is warm and resonant. The spatial environment remains a small, untreated room with audible, short reverberation
loudness: moderate
10.96–11.28s speech
speaker speaker_0
No.
♥ neutral, definitive
♫ conversational, brief response
adult_30s · female · en · American English · normal · mezzo-soprano, clear
11.28–12.6s sound timeline fill (SFX fallback)
The same quiet background hiss and room tone are present beneath the speech. The vocal timbre is warm and resonant. The spatial environment remains a small, untreated room with audible, short reverberation
loudness: moderate
12.6–14.6s sound
loud, sharp crash/shattering sound
loudness: very_loud
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: Well, to me, how did it sound to you? Well, I, I, I wouldn't know. Is that an unusual sound for a chimp? No. [Noise]
Parakeet: How did it sound to you? Well, I I mean I I wouldn't know. Is that an unusual sound for a chimp? No.
Qwen3-ASR: To me how did it sound to you Well I wouldn't know Is that an unusual sound for a chimp No
Raw SFX LoRA detections (3) — fed into MOSS as candidate sound events
0.0–7.8s An adult female with a low-pitched, slightly breathy voice speaks with an emotional, questioning, and strained tone. Her delivery is slow and deliberate, with a sense of vulnerability. The recording is captured at a medium proximity in what sounds like a small, untreated room. Throughout the segment, a quiet but noticeable background hiss and room tone are audible. Universal Technical Audio Analysis: The overall loudness is quiet to conversational, with natural dynamic variations in the speech. The spectral balance is focused on the mid-range, with a touch of low-mid warmth, lacking significant high-frequency air. The voice has a soft, slightly airy texture. The spatial acoustics suggest a small, slightly live room with short, natural reverberation. The recording sounds unprocessed, with the prominent room tone and hiss indicating a raw or ambient capture rather than a clean studio recording. There is clear separation between the voice and the quiet noise floor.
7.8–15.3s An adult male with a mid-range, resonant voice responds in a conversational, slightly hesitant manner that sounds reassuring. He speaks at a closer proximity and at a steady, moderate pace. The same quiet background hiss and room tone are present beneath the speech. Universal Technical Audio Analysis: The loudness is at a consistent conversational level. The voice has a fuller body in the low-mids due to the closer proximity, with clear mid-range presence. The vocal timbre is warm and resonant. The segment begins with a hard cut following the end of the previous speaker's utterance. The spatial environment remains a small, untreated room with audible, short reverberation. The audio is presented without any apparent compression, mixing, or effects, preserving the raw quality of the voice and the ambient noise.
12.6–14.6s A sudden, extremely loud, and sharp sound of shattering glass erupts, dominating the audio. The sound has a violent attack with a wide frequency spectrum, rich in high-frequency transients. It has a brief but intense reverberant tail that suggests a small, reflective space. Universal Technical Audio Analysis: The event is exceptionally loud, pushing the recording into heavy digital clipping and auditory saturation. This limits the available dynamic range and creates harsh, distorted artifacts. The spectral balance is extremely bright and aggressive, with a piercing high-end. The texture is brittle, glassy, and fragmented. Spatially, the sound feels wide but is located centrally, with a fast decay and a noticeable reverberant tail. The onset is instantaneous and explosive, and the segment concludes with the decay of the reverberation's early reflections.

love_potion_monkey_test_clip03

Love Potion No. 9 - Monkey Test.mp4 · 57.775s–106.099s · 48.353s
timetypeannotation
0.0–6.8s sound timeline fill (SFX fallback)
A violent, chaotic struggle begins in a reverberant room with a heavy metallic clatter and latch sound. This is immediately followed by a sequence of loud, sharp, metallic impacts, including a distinct door slam. Concurrently, a loud door slamming sound echoes. Universal Technical Audio Analysis: The overall level is very loud with explosive, sharp transients. The spectral balance is bright and aggressive, dominated by the high-frequency content of the metallic impacts. The timbre of the impacts is hard and resonant. The spatial acoustics suggest a medium-sized room with hard surfaces, creating a noticeable, bright reverberation. Sources are localized centrally. The segment begins with a hard cut into the action
loudness: moderate
6.8–8.24s sound timeline fill (SFX fallback)
In the background, intermittent distant gunshots from a different location echo a series of loud, metallic impacts from a close, violent struggle continue persistently. the background sounds have a prominent, long reverb tail, clearly establishing spatial depth
loudness: moderate
8.24–10.24s speech
speaker speaker_0
Romeo he's trying to break through the wall What's he trying to do He's trying to break
♥ clearly panicked and distressed
♫ urgent, pleading, fast-paced
young_adult_20s · male · en · American · fast · moderately_bright, slightly_tense, clear
10.24–14.0s sound timeline fill (SFX fallback)
In the background, intermittent distant gunshots from a different location echo a series of loud, metallic impacts from a close, violent struggle continue persistently. the background sounds have a prominent, long reverb tail, clearly establishing spatial depth
loudness: moderate
14.0–27.8s sound timeline fill (SFX fallback)
A continuous and intense firefight erupts with multiple loud, sharp reports of gunfire, heavily colored by the reverberation of the environment. Overlapping the gunfire are prominent metallic impact sounds, like bullet ricochets or objects being hit. A high-pitched scream punctuates the chaos at two points. The segment ends abruptly with a cut. Universal Technical Audio Analysis: The loudness is consistently high and heavily compressed, with very little dynamic variation. The spectral balance is harsh and aggressive, dominated by the mid-range and high frequencies of the gunshots and ricochets. The timbre is sharp, percussive, and metallic. The spatial acoustics feature a large reverberation, creating a sense of a large, open, but reflective space
loudness: moderate
27.8–48.0s sound timeline fill (SFX fallback)
A complex and dense sonic layer begins with the sound of a powerful, rhythmic electronic music track. This features a heavy kick drum, a deep sub-bass line, synthesized percussion, and an aggressive, processed male vocal sample. This music track is layered over the continuing sounds of gunfire and impacts from a chaotic scene. The vocal sample repeats with a driving cadence. Universal Technical Audio Analysis: The overall level is very loud and heavily limited, creating a dense wall of sound. The spectral balance is full-range, with the music providing a heavy low-end foundation the gunfire and vocals occupy the mid and high frequencies. The timbre of the music is synthetic and gritty the gunshots remain sharp and metallic. The heavy reverb from the chaos scene and the music's processing create a fused, saturated spatial environment. There is significant masking occurring between the musical and sound effects layers
loudness: moderate
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Music]
Parakeet: —
Qwen3-ASR: Romeo he's trying to break through the wall What's he trying to do He's trying to break through the wall
Raw SFX LoRA detections (4) — fed into MOSS as candidate sound events
0.0–6.8s A violent, chaotic struggle begins in a reverberant room with a heavy metallic clatter and latch sound. This is immediately followed by a sequence of loud, sharp, metallic impacts, including a distinct door slam. A female voice shouts in a panicked, high-pitched tone. Concurrently, a loud door slamming sound echoes. Universal Technical Audio Analysis: The overall level is very loud with explosive, sharp transients. The spectral balance is bright and aggressive, dominated by the high-frequency content of the metallic impacts. The timbre of the impacts is hard and resonant. The spatial acoustics suggest a medium-sized room with hard surfaces, creating a noticeable, bright reverberation. Sources are localized centrally. The segment begins with a hard cut into the action.
6.8–14.0s A male voice with an urgent, pleading tone speaks at a conversational level. His voice is clear and recorded with close proximity, contrasting with the surrounding chaos. In the background, intermittent distant gunshots from a different location echo, while a series of loud, metallic impacts from a close, violent struggle continue persistently. Universal Technical Audio Analysis: The segment has a wide dynamic range, with the controlled, mid-range focused male voice juxtaposed against loud, broadband background events. The male voice has a dry tonal quality with little reverb, while the background sounds have a prominent, long reverb tail, clearly establishing spatial depth. The foreground voice appears compressed for clarity against the dynamic scene.
14.0–27.8s A continuous and intense firefight erupts with multiple loud, sharp reports of gunfire, heavily colored by the reverberation of the environment. Overlapping the gunfire are prominent metallic impact sounds, like bullet ricochets or objects being hit. A high-pitched scream punctuates the chaos at two points. The segment ends abruptly with a cut. Universal Technical Audio Analysis: The loudness is consistently high and heavily compressed, with very little dynamic variation. The spectral balance is harsh and aggressive, dominated by the mid-range and high frequencies of the gunshots and ricochets. The timbre is sharp, percussive, and metallic. The spatial acoustics feature a large reverberation, creating a sense of a large, open, but reflective space.
27.8–48.0s A complex and dense sonic layer begins with the sound of a powerful, rhythmic electronic music track. This features a heavy kick drum, a deep sub-bass line, synthesized percussion, and an aggressive, processed male vocal sample. This music track is layered over the continuing sounds of gunfire and impacts from a chaotic scene. The vocal sample repeats with a driving cadence. Universal Technical Audio Analysis: The overall level is very loud and heavily limited, creating a dense wall of sound. The spectral balance is full-range, with the music providing a heavy low-end foundation while the gunfire and vocals occupy the mid and high frequencies. The timbre of the music is synthetic and gritty, while the gunshots remain sharp and metallic. The heavy reverb from the chaos scene and the music's processing create a fused, saturated spatial environment. There is significant masking occurring between the musical and sound effects layers.

love_potion_monkey_test_clip04

Love Potion No. 9 - Monkey Test.mp4 · 106.099s–139.442s · 33.358s
timetypeannotation
0.0–9.0s sound
Loud rock music intro with distorted guitars, heavy drums, and sound effects like a vinyl scratch and metallic shifting. The music cuts off abruptly.
loudness: loud
8.5–18.5s sound
Very quiet room tone with subtle, intermittent rustling sounds, like fabric movement. A faint electronic hiss is present.
loudness: quiet
18.5–19.76s sound timeline fill (SFX fallback)
near silence with faint room tone
loudness: quiet
19.76–21.04s speech
speaker speaker_0
Is he dead?
♥ neutral
♫ casual, questioning
adult_30s · male · en · American English · normal · medium-pitched, clear, slightly soft
21.04–26.16s sound timeline fill (SFX fallback)
A quiet room tone is audible underneath the speech. The spatial acoustics are very dry and intimate, with minimal room reverberation, creating a close-mic feel. The recording is clean, with the only other audible element being the low-level ambient room tone
loudness: moderate
26.16–26.48s speech
speaker speaker_1
No.
♥ neutral, slightly soft
♫ soft, calm
adult_30s · female · en · American English · normal · high-pitched, clear, slightly breathy
26.48–27.12s sound timeline fill (SFX fallback)
A quiet room tone is audible underneath the speech. The spatial acoustics are very dry and intimate, with minimal room reverberation, creating a close-mic feel. The recording is clean, with the only other audible element being the low-level ambient room tone
loudness: moderate
27.12–27.48s vocal burst
speaker speaker_1
soft laugh
♥ amused, relieved
27.5–33.5s sound
Quiet room tone with faint background noise.
loudness: quiet
30.64–32.16s speech
speaker speaker_1
He's asleep.
♥ relieved, soft
♫ soft, calm
adult_30s · female · en · American English · normal · high-pitched, clear, slightly breathy
Raw transcripts from the three ASR systems (before reconciliation)
VibeVoice: [Music] Is he dead? [Environmental Sounds] No. [Human Sounds] He's asleep.
Parakeet: Is he dead? No. He's asleep.
Qwen3-ASR: Is he dead No He's asleep
Raw SFX LoRA detections (4) — fed into MOSS as candidate sound events
0.0–9.0s A loud, aggressive rock music intro is playing, featuring distorted electric guitars, a driving drum beat with a prominent cymbal crash, and a gritty bassline. The intense music is accompanied by sound effects, including a vinyl record scratch near the beginning, the sound of a heavy, metallic object shifting, and rustling noises. The music comes to an abrupt stop partway through the segment, but the mechanical and rustling sounds continue briefly before fading out. Universal Technical Audio Analysis: The segment is very loud and heavily compressed, with a full-range spectral balance dominated by the mid-frequencies of the distorted guitars. The tonal character is gritty and aggressive. The spatial presentation is wide but lacks a strong sense of room acoustics, feeling more direct. The offset is a hard, abrupt cut. The mix is dense, with the initial transients of the drums and effects punching through the wall of guitar distortion.
8.5–18.5s A period of very quiet, subtle room tone is audible, characterized by intermittent, very soft rustling sounds, akin to gentle fabric movement or slight body shifts. The movement is slow and sparse, suggesting minimal activity in a quiet indoor space. A faint, steady electronic hiss or noise floor is perceptible in the background throughout this quiet passage. Universal Technical Audio Analysis: The overall level is very quiet, with a dynamic range defined by the low-level rustles against the near-silent base. The spectral content is concentrated in the low-mids and high frequencies of the rustling texture. The acoustic character of the space is dry, suggesting a small, non-reverberant room. There is no obvious processing; the sound is direct and uncolored, capturing natural, low-level sounds with a close-proximity feel.
19.5–28.0s A conversational exchange occurs in a quiet setting. An adult male voice asks a brief, quiet question, followed by an adult female voice responding in a hushed, calm, and slightly breathy tone. The female voice is positioned slightly more centrally and appears closer to the microphone than the male voice. A quiet room tone is audible underneath the speech. Universal Technical Audio Analysis: The loudness is at a conversational level, with natural dynamics in the speech, though the delivery is deliberately quiet. The tonal balance is natural, with the female voice exhibiting more breathiness in its high-frequency range. The spatial acoustics are very dry and intimate, with minimal room reverberation, creating a close-mic feel. The recording is clean, with the only other audible element being the low-level ambient room tone.
27.5–33.5s An adult female voice speaks in a hushed, breathy, and calm tone. After a brief pause, she laughs very quietly, almost into a sigh, continuing to speak in the same quiet manner. The delivery is slow and deliberate, suggesting a conspiratorial or secretive atmosphere within a quiet indoor space. A faint, steady room tone is present in the background. Universal Technical Audio Analysis: The speech is quiet and dynamically controlled, with soft onsets for both words and the subsequent laugh. The spectral balance is natural, with the breathy quality of the female voice adding airiness in the higher frequencies. The spatial impression is very intimate and dry, consistent with a close microphone in a small room. The recording is clean, capturing the subtle textures of the breathy speech without significant coloration or processing.

Generated from 10 test clips · triple_greedy config.