vocal-burst
synthetic_vocal_burstsThis repository contains the vocal bursts like giggling, laughter, shouting, crying, etc. from the following repository.
https://huggingface.co/datasets/sleeping-ai/Vocal-burst
We captioned them using Gemini Flash Audio 2.0. This dataset contains, this dataset contains ~ 365,000 vocal bursts from all kinds of categories.
It might be helpful for pre-training audio text foundation models to generate and understand all kinds of nuances in vocal bursts.
vocal-bursts-segments
vocal-bursts-segments
128,165 vocal-burst segments from two acoustically unrelated sources, cut with one policy, plus
46,494 verified no-burst segments. Every segment is a single burst — a laugh, a
sigh, a gasp — and nothing else.
subtree
what it is
segments
real/
real speech (Emilia, LAION voice profiles, vocal-bursts-clean, Kartoffelphon), the 5,161 segments of laion/vocal-bursts-gemini-segments
5,161
dramabox/
synthetic voice-acting output, cut out of… See the full description on the dataset page: https://huggingface.co/datasets/laion/vocal-bursts-segments.vocal-bursts-clean
Vocal Bursts — Clean
2,285,651 isolated non-verbal vocal bursts cut from in-the-wild expressive speech,
each with a single class label. 284.77 hours of audio,
68 classes, 20.7 GiB.
Two columns, nothing else:
column
type
content
mp3
audio
the burst, 160 kbps mono MP3
txt
string
the class label, e.g. Sharp Inhale
This is the filtered release. Everything the pipeline flagged as unreliable has already
been removed — see How it was filtered. If you want the raw… See the full description on the dataset page: https://huggingface.co/datasets/laion/vocal-bursts-clean.vocal-bursts-gemini-segments
burst_gemini_segments (Dataset B)
5,161 vocal-burst segments cut out of 3,304 real speech utterances, one per
event that Gemini 3.8 Flash asserted. Each file is a single burst — a laugh, a sigh, a gasp — and
nothing else. Median length 0.76 s; 1.39 hours in total.
Built to retrain a burst classifier. The detector this project shipped emitted Shriek zero
times over a 60-clip audit, used 8 of its 83 labels, and put the requested burst in its top-3 on
3 of 60 clips. Every burst… See the full description on the dataset page: https://huggingface.co/datasets/laion/vocal-bursts-gemini-segments.vocal-bursts-per-class
Vocal bursts, per class
33,991 clips over 45 vocal-burst classes, one loadable config per class:
from datasets import load_dataset
ds = load_dataset("laion/vocal-bursts-per-class", "scream")
Every clip carries a gemini-3.8-flash burst annotation -- spans, 1-3 labels most-likely-first,
a confidence and a free-text description -- and the rank at which it qualified for its class.
Two halves, never mixed silently:
half
clips
what it is
real
3,314
real recorded or… See the full description on the dataset page: https://huggingface.co/datasets/laion/vocal-bursts-per-class.vocal-burst-classification-v2
Vocal Burst Classification V2 — laion/vocal-burst-classification-v2
The V2 training corpus for the Vocal Burst Classifier V2:
a single-label dataset over an 83-class vocal-burst taxonomy (82 non-speech human vocalizations
no_burst, index 82), shipped as precomputed VoiceCLAP-commercial embeddings plus the raw
vocal-bursts-clean audio.
The vocal-burst clips were generated with various synthetic text-to-audio models such as DramaBox,
then annotated and filtered as described… See the full description on the dataset page: https://huggingface.co/datasets/laion/vocal-burst-classification-v2.