Speech-free & music-free source clips — a random sample of 12

Drawn at random (seed 20260809) from the 8,191 clips that survived a two-stage audio filter over 101,503 source videos from InsAVE-80K and JAVEdit-100k: no speech (FireRedVAD, then FireRedASR2-AED confirmation on every VAD positive) and no music-related AudioSet tag anywhere in the top five. These are the inputs to the removal pipeline, not its outputs. Play with sound on.

101,503 clips screened
31,845 speech-free
8,191 also music-free
12 sampled at random, not curated
Sampled, not cherry-picked — so the failure mode is visible too. 3 of these 12 carry Speech as their top AudioSet tag even though VAD and ASR both cleared them. That is the known model disagreement: FireRedASR2 covers Chinese and English, so speech in another language transcribes empty and slips through. Those clips are flagged in the manifest via speech_disagreement, and the stricter variant (kept_no_speech_strict.csv, 6,983 clips) drops them.
Sheep InsAVE-80K general_editing
1280×704 · 24 fps · 5.043s · mp3 44kHz stereo
insave_general_editing_48600
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Sheep 0.67Bleat 0.59Animal 0.57Goat 0.55Livestock, farm animals, working animals 0.20
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0066audio RMS
Animal InsAVE-80K eval
1280×704 · 23.98 fps · 3.397s · mp3 44kHz stereo
insave_eval_00231
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Animal 0.12Speech 0.11Door 0.11Bird 0.07Chicken, rooster 0.06
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0161audio RMS
SpeechAudioSet says speech InsAVE-80K add_and_remove
1280×704 · 24 fps · 3.397s · mp3 44kHz stereo
insave_add_and_remove_04959
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Speech 0.37Animal 0.07Sneeze 0.06Gasp 0.04Clip-clop 0.03
0.96svoice activity detected
yesASR confirmation
—transcript (empty = no speech)
0.0069audio RMS
SpeechAudioSet says speech JAVEdit-100k source
1280×720 · 25 fps · 4.864s · aac 16kHz mono
javedit_7162ff62de8f
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Speech 0.81Silence 0.08Television 0.06Male speech, man speaking 0.05Inside, small room 0.03
1.23svoice activity detected
yesASR confirmation
—transcript (empty = no speech)
0.0049audio RMS
SpeechAudioSet says speech InsAVE-80K general_editing
1280×704 · 23.98 fps · 6.715s · mp3 44kHz stereo
insave_general_editing_34794
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Speech 0.66Animal 0.28Bird 0.10Duck 0.05Quack 0.05
2.18svoice activity detected
yesASR confirmation
—transcript (empty = no speech)
0.0092audio RMS
Door InsAVE-80K add_and_remove
1280×704 · 23.98 fps · 5.069s · mp3 44kHz stereo
insave_add_and_remove_07472
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Door 0.30Slam 0.17Walk, footsteps 0.09Thunk 0.08Sliding door 0.05
0.61svoice activity detected
yesASR confirmation
—transcript (empty = no speech)
0.0206audio RMS
Liquid InsAVE-80K general_editing
1280×704 · 23.98 fps · 3.397s · mp3 44kHz stereo
insave_general_editing_31165
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Liquid 0.15Rowboat, canoe, kayak 0.11Boat, Water vehicle 0.09Stream 0.08Animal 0.08
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0029audio RMS
Frog InsAVE-80K general_editing
1280×704 · 23.98 fps · 3.397s · mp3 44kHz stereo
insave_general_editing_47918
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Frog 0.30Croak 0.11Animal 0.05Walk, footsteps 0.04Outside, rural or natural 0.04
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0019audio RMS
Typewriter InsAVE-80K add_and_remove
1280×704 · 23.98 fps · 6.715s · mp3 44kHz stereo
insave_add_and_remove_04160
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Typewriter 0.28Door 0.27Sliding door 0.12Printer 0.04Cash register 0.04
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0018audio RMS
Clip-clop JAVEdit-100k source
1280×720 · 25 fps · 4.864s · aac 16kHz mono
javedit_5b6a0a248356
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Clip-clop 0.63Animal 0.57Horse 0.40Walk, footsteps 0.10Outside, rural or natural 0.08
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0024audio RMS
Vehicle JAVEdit-100k source
1280×720 · 25 fps · 4.864s · aac 16kHz mono
javedit_4830e7822b49
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Vehicle 0.52Car 0.13Truck 0.07Medium engine (mid frequency) 0.06Bus 0.05
0svoice activity detected
not neededASR confirmation
—transcript (empty = no speech)
0.0545audio RMS
Animal InsAVE-80K add_and_remove
1280×704 · 23.98 fps · 3.397s · mp3 44kHz stereo
insave_add_and_remove_05404
frames
6 frames, evenly spaced
spectrogram
audio spectrogram (0–8 kHz)
Animal 0.39Domestic animals, pets 0.27Bow-wow 0.26Dog 0.21Speech 0.15
0.53svoice activity detected
yesASR confirmation
—transcript (empty = no speech)
0.0332audio RMS