datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
BEAT2MRSAudio
MRSAudio: A Large-Scale Multimodal Recorded Spatial Audio Dataset with Refined Annotations
Humans rely on multisensory integration to perceive spatial environments, where auditory cues enable sound source localization in three-dimensional space.
Despite the critical role of spatial audio in immersive technologies such as VR/AR, most existing multimodal datasets provide only monaural audio, which limits the development of spatial audio generation and understanding.
To address… See the full description on the dataset page: https://huggingface.co/datasets/verstar/MRSAudio.AV-SpeakerBench
AV-SpeakerBench
Audiovisual QA benchmark with speaker-aware questions and aligned clips. This drop includes trimmed segments (audio-only, visual-only, audiovisual) plus annotations to probe fine-grained AV reasoning.
Project page: https://plnguyen2908.github.io/AV-SpeakerBench-project-page/
Code & benchmarks: https://github.com/plnguyen2908/AV-SpeakerBench
Paper: https://arxiv.org/abs/2512.02231
Files
test.csv - original annotations and metadata with clip paths… See the full description on the dataset page: https://huggingface.co/datasets/plnguyen2908/AV-SpeakerBench.WildElder
WILDELDER: A CHINESE ELDERLY SPEECH DATASET FROM THE WILD WITH FINE-GRAINED MANUAL ANNOTATIONS
Paper: https://huggingface.co/papers/2510.09344Code: https://github.com/NKU-HLT/WildElder
WildElder is a speech dataset focused on elderly scenarios. It contains raw audio and corresponding text annotations and can be used for ASR, speaker-related tasks, and front-/back-end speech processing research. The data was collected and cleaned from real-world environments to preserve diversity and… See the full description on the dataset page: https://huggingface.co/datasets/Hui519/WildElder.beat2-additional-annotations
BEAT2 Official Release + Additional Annotations
This is a fork of H-Liu1997/BEAT2
that adds annotations contributed by the
RAG-Gesture (CVPR 2025)
and MIBURI (CVPR 2026) projects.
The base BEAT2-English data (motion, audio, TextGrids, semantic labels,
pretrained motion-autoencoder weights) is inherited verbatim from upstream;
the additional annotations from RAG-Gesture and MIBURI are pushed on top.
Citations
If you use only the original BEAT2 dataset, please cite… See the full description on the dataset page: https://huggingface.co/datasets/m-hamza-mughal/beat2-additional-annotations.AudioMarathon
🎵 AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficient Inference in Multimodal LLMs
Abstract
AudioMarathon is a large-scale, multi-task audio understanding benchmark designed to systematically evaluate audio language models' capabilities in processing and comprehending long-form audio content. It provides a diverse set of 10 tasks built upon three pillars:
long-context audio inputs with durations ranging from 90.0 to 300.0… See the full description on the dataset page: https://huggingface.co/datasets/Hezep/AudioMarathon.wikimedia-common-audio-catalanThis is a collection of Catalan-language audio with free licenses extracted from Wikimedia Commons.
License identifiers are normalized to cc-zero, cc-by-4.0,
cc-by-sa-3.0, cc-by-sa-4.0, GFDL, and PD-self.
This provides a richer alternative to Common Voice.
Characteristics of the dataset:
One or multiple speakers
Different accents
Different domain texts
761 audio files
We found this dataset useful for audio tasks such as:
Language detection
Evaluation of STT systems
New candidates are… See the full description on the dataset page: https://huggingface.co/datasets/softcatala/wikimedia-common-audio-catalan.UrbanSound8K
UrbanSound8K
This is an audio classification dataset for Sound Event Classification.
Classes = 10 , Split = Ten-Fold
Structure
audios folder contains audio files.
csv_files folder contains CSV files for ten-fold cross-validation.
To perform cross-validation on fold 1, train_1.csv will be used for the training split and test_1.csv for the testing split, with the same pattern followed for the other folds.
To perform training and testing witout cross-validation, use… See the full description on the dataset page: https://huggingface.co/datasets/MahiA/UrbanSound8K.Raon-OpenTTS-Eval
Raon-OpenTTS-Eval
Technical Report
A robustness-oriented evaluation benchmark for zero-shot text-to-speech, covering 4 acoustic regimes (Clean, Noisy, Wild, Expressive) across 12 datasets with 6,000 prompt–text pairs.
Existing zero-shot TTS benchmarks typically evaluate models using prompts drawn from a single read-speech dataset, providing an incomplete view of robustness under realistic and challenging recording scenarios. Raon-OpenTTS-Eval… See the full description on the dataset page: https://huggingface.co/datasets/KRAFTON/Raon-OpenTTS-Eval.ahead_ds
Another HEaring AiD DataSet (AHEAD-DS)
Another HEaring AiD DataSet (AHEAD-DS) is an audio dataset labelled with audiologically relevant scene categories for hearing aids.
Website
Paper
Code
Dataset AHEAD-DS
Dataset AHEAD-DS unmixed
Models
Description of data
All files are encoded as single channel WAV, 16 bit signed, sampled at 16 kHz with 10 seconds per recording.
Category
Training
Validation
Testing
All
cocktail_party
934
134
266
1334
interfering_speakers… See the full description on the dataset page: https://huggingface.co/datasets/hzhongresearch/ahead_ds.suno-ai-music-dataset
Suno AI Music Dataset (Multi-Genre Curated)
A human-curated, multi-genre audio dataset generated with Suno V5.5 (chirp-fenix), covering 100+ sub-sub-genres across electronic, hip-hop, Latin, jazz, world, rock, ambient, pop, reggae, and classical music. Each track ships with full audio (MP3), cover art, the original generation prompt, and a 32-column metadata schema designed for downstream audio-ML research.
This is not a "scrape everything Suno produces" dump. It is a… See the full description on the dataset page: https://huggingface.co/datasets/Kukedlc/suno-ai-music-dataset.MANGO
MANGO: A Corpus of Human Ratings for Speech
MANGO (MUSHRA Assessment corpus using Native listeners and Guidelines to understand human Opinions at scale) is the first large-scale dataset designed for evaluating Text-to-Speech (TTS) systems in Indian languages.
Key Features:
255,150 human ratings of TTS-generated outputs and ground-truth human speech.
Covers two major Indian languages: Hindi & Tamil, and English.
Based on the MUSHRA (Multiple Stimuli with Hidden Reference… See the full description on the dataset page: https://huggingface.co/datasets/ai4bharat/MANGO.the-truthChinese-LiPS
Chinese-LiPS: A Chinese audio-visual speech recognition dataset with Lip-reading and Presentation Slides
⭐ Introduction
The Chinese-LiPS dataset is a multimodal dataset designed for audio-visual speech recognition (AVSR) in Mandarin Chinese. This dataset combines speech, video, and textual transcriptions to enhance automatic speech recognition (ASR) performance, especially in educational and instructional scenarios.
🚀 Dataset Details
Total Duration:… See the full description on the dataset page: https://huggingface.co/datasets/BAAI/Chinese-LiPS.PianoVAM_v1
PianoVAM v1.2: A Multimodal Piano Performance Dataset
Version History
v1.2 (current). Adds Fingering/ (per-note fingering labels for 106 recordings) and Fingering_GT/ (manual fingering annotations for 11 recordings). All other files are unchanged from v1.1.
v1.1. metadata.json is the canonical split file. Video files for the 'Sep 04-05' recordings are the sync-corrected versions, and files previously found to have video-MIDI synchronization issues have been… See the full description on the dataset page: https://huggingface.co/datasets/PianoVAM/PianoVAM_v1.Nemotron-Content-Safety-Audio-Dataset
Nemotron Content Safety Audio Dataset
Dataset Description
The Nemotron Content Safety Audio Dataset is a multimodal extension of the Nemotron Content Safety Dataset V2 (Aegis 2.0), comprising 1,928 audio files generated from the test set prompts. This dataset enables multimodal AI safety research by providing spoken versions of adversarial and safety-critical prompts across 23 violation categories.
LANGUAGE: All prompts are in English. However, the audio files were… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Content-Safety-Audio-Dataset.sada2022
Dataset Card for SADA صدى
Dataset Summary
يعتبر توفر البيانات من أهم ممكنات تطوير نماذج ذكاء اصطناعي متفوقة إن لم يكن أهمها، ولكن لا تزال البيانات الصوتية المفتوحة وخصوصاً باللغة العربية ولهجاتها المختلفة شحيحة المصدر.
ومن هذا المنطلق وحرصًا على إطلاق القيمة الكامنة للبيانات وتمكين تطوير منتجات مبنية على الذكاء الاصطناعي، قام المركز الوطني للذكاء الاصطناعي في سدايا (الهيئة الوطنية للبيانات والذكاء الاصطناعي) بالتعاون مع الهيئة السعودية للإذاعة والتلفزيون بنشر مجموعة… See the full description on the dataset page: https://huggingface.co/datasets/khaledalganem/sada2022.asr-jargon-specialized-vocabulary
A Dataset for Evaluating ASR on Specialized Vocabulary
Novel synthetic datasets from the paper "A Dataset for Evaluating ASR on Specialized Vocabulary" (LREC 2026).
Code and reproduction scripts: https://github.com/eduardogc8/ASR-Jargon-Dataset-Code
Configs
Config
Language
Description
synthetic_terms_en
English
Utterances embedding entirely novel, 100% OOV, LLM-generated technical terms
synthetic_terms_pt
Portuguese
Portuguese equivalent… See the full description on the dataset page: https://huggingface.co/datasets/egcortes/asr-jargon-specialized-vocabulary.AudioMarathon
AudioMarathon
AudioMarathon is a long-context audio benchmark for evaluating multimodal LLMs on speech, music, environmental audio, and meetings. The release package in this directory is organized around 11 benchmark tasks spanning meeting summarization, automatic speech recognition, reading comprehension, authenticity detection, music genre classification, acoustic scene classification, emotion recognition, spoken named entity reasoning, sound event detection, speaker gender… See the full description on the dataset page: https://huggingface.co/datasets/AudioMarathon/AudioMarathon.ArFake
ArFake-Dataset
ARFAKE: A Robust Framework for Multi-Dialect Arabic Speech Spoofing Detection Benchmark
ARFAKE is the first end-to-end benchmark for Arabic speech spoofing detection across multiple dialects. The framework systematically generates synthetic Arabic speech, evaluates its intelligibility and realism, constructs a large-scale spoofing dataset, trains robust detectors, and evaluates generalization across both unseen generators and unseen dialects.… See the full description on the dataset page: https://huggingface.co/datasets/Mohammed01/ArFake.QualiSpeech
QualiSpeech: A Speech Quality Assessment Dataset with Natural Language Reasoning and Descriptions
📄 Paper: https://arxiv.org/abs/2503.20290
QualiSpeech is a comprehensive English-language speech quality assessment dataset designed to go beyond traditional numerical scores. It introduces detailed natural language comments with reasoning, capturing low-level speech perception aspects such as noise, distortion, continuity, speed, naturalness, listening effort, and overall… See the full description on the dataset page: https://huggingface.co/datasets/tsinghua-ee/QualiSpeech.SynthGT
SynthGT
A Synthetic Solo-Singing Dataset for Singing-Oriented Forced Alignment
Authors
Silas Antonisen, Iván López-Espejo
Associated paper submitted to IEEE Transactions on Audio, Speech and Language Processing.
Overview
SynthGT (Synthetic Ground Truth) is a synthetic English solo-singing dataset containing 4,900 singing performances with automatically generated phoneme boundary annotations.
The dataset was created through music… See the full description on the dataset page: https://huggingface.co/datasets/Silasimo/SynthGT.TaxaBench-8k
Paper: TaxaBind: A Unified Embedding Space for Ecological Applications
Venue: WACV 2025
Github: https://github.com/mvrl/TaxaBind
Dataset Name: TaxaBench-8k
Dataset Description:
TaxaBench-8k is a multimodal dataset containing six modalities - image, text, satellite image, audio, geographic location, and environmental features for evaluating large ecological models.
Usage:
Please use the test_df.csv for reading data which… See the full description on the dataset page: https://huggingface.co/datasets/MVRL/TaxaBench-8k.doppelganger
Doppelganger: Sound Effects and Their Synthetic Twins
Benchmark for matching a synthetic sound effect to the real recording it was generated from.
Paper: https://arxiv.org/abs/2607.04337 · Code: https://github.com/elliottash/doppelganger · models:
https://huggingface.co/elliottash/doppelganger
Contents
sao_pairs/<CatID>/<instance_id>.wav — Stable-Audio-Open audio-conditioned synthetic twins (the
main UCS corpus, one twin per verified real clip).… See the full description on the dataset page: https://huggingface.co/datasets/elliottash/doppelganger.strikerData
🎧 StrikerData
Overview
StrikerData is an audio dataset developed by Strikersoft for research and development in audio and speech technologies.It contains human speech, environmental noise, and other sound types. The dataset is available for non-commercial use only, except for the company Strikersoft.
Category
Percentage of Total Dataset
Clean human speech
20%
Distorted speech
15%
Human-made noise
15%
Non-human noise
50%
⚖️ License… See the full description on the dataset page: https://huggingface.co/datasets/strikersoft/strikerData.Neapolitan-Spoken-Corpus
Neapolitan Spoken Corpus (NSC)
A corpus of read Neapolitan speech for ASR evaluation, with a validated
Neapolitan–Italian lexicon, LOSO fine-tuning splits, trained LoRA adapters,
metric implementations, per-clip results, and error annotations.
This release supersedes the earlier 141-clip single-speaker version of this
repository. The earlier release corresponds to Speaker S1 of the present
corpus; the old audioData/ and transcripts.csv are replaced by
data/audio/ and… See the full description on the dataset page: https://huggingface.co/datasets/michaelcacioli/Neapolitan-Spoken-Corpus.serena-synthetic-it-28h
Qwen3-TTS Italian Synthetic Speech (27h)
Synthetic Italian single-speaker speech dataset for TTS training (e.g. Piper), generated with
Qwen3-TTS-1.7B-Base in voice-cloning mode. ~29.5k clips, ~27 hours, 22.05 kHz mono WAV,
Piper-ready metadata.
Dataset summary
Property
Value
Clips (train / val)
26,523 / 2,947
Total duration
~27.3 h (98,099 s)
Sample rate
22,050 Hz mono, 16-bit WAV
Loudness
Normalized to -23 LUFS, silence-trimmed
Language
Italian… See the full description on the dataset page: https://huggingface.co/datasets/committa/serena-synthetic-it-28h.CREMA-D
CREMA-D
This is an audio classification dataset for Emotion Recognition.
Classes = 6 , Split = Train-Test
Structure
audios folder contains audio files.
train.csv for training split and test.csv for the testing split.
Download
import os
import huggingface_hub
audio_datasets_path = "DATASET_PATH/Audio-Datasets"
if not os.path.exists(audio_datasets_path): print(f"Given {audio_datasets_path=} does not exist. Specify a valid path ending with 'Audio-Datasets'… See the full description on the dataset page: https://huggingface.co/datasets/MahiA/CREMA-D.EK100
Motivation
The actual download link is very slow, including the academic torrent. Therefore, to spare fellow community members from this misery, I am uploading the dataset here.
Source
You can fnd the original source to download the dataset: https://github.com/epic-kitchens/epic-kitchens-download-scripts
Citation
@INPROCEEDINGS{Damen2018EPICKITCHENS,
title={Scaling Egocentric Vision: The EPIC-KITCHENS Dataset},
author={Damen, Dima and Doughty, Hazel and… See the full description on the dataset page: https://huggingface.co/datasets/itruonghai/EK100.MADB-Dataset
MADB: Music Aesthetics Dataset and Benchmark
Dataset Description
MADB is a large-scale dataset for music aesthetic evaluation, designed to support research on multi-dimensional and subjective music perception.
The dataset contains approximately 10,000 music tracks, each annotated by multiple trained annotators across 10 perceptual dimensions and one overall score. In addition, each track includes textual comments and semantic tags (genre and mood), enabling… See the full description on the dataset page: https://huggingface.co/datasets/sirui1/MADB-Dataset.