datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
omnivoice-best-of-n-training
🎙 Best-of-N Voice Cloning Training Data
Curated training dataset for fine-tuning OmniVoice for IWSLT 2026.
🎧 Listen to the Audio
This dataset has playable audio columns — click on any row in the dataset viewer
to listen to both the reference audio (original speaker) and the best synthesized audio
(selected by quality score).
Dataset Description
For each sentence in the dev split of ymoslem/acl-6060 (468 samples × 3 languages),
we synthesized audio with the… See the full description on the dataset page: https://huggingface.co/datasets/amanuelbyte/omnivoice-best-of-n-training.bashkort_commands_omnivoice
Bashkort Commands OmniVoice
Partial eleven-label command snapshot generated with k2-fsa/OmniVoice
using the same cross-lingual voice-cloning recipe as
AigizK/homai_wake_word_omnivoice. Generation was stopped at the user's
request after 41,525 complete reference groups had been committed.
For every included reference row from the train split of:
bond005/sova_rudevices
the dataset contains one recording of every command:
Айвика — Russian
Айвикә — Bashkir
Айһылыу — Bashkir… See the full description on the dataset page: https://huggingface.co/datasets/AigizK/bashkort_commands_omnivoice.omnivoice-tr
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
10,000
Total
20,000
omnivoice-fr
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
10,000
Total
20,000
omnivoice-th
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
9,833
Total
19,833
msa-omnivoice-tts-v1
MSA-OmniVoice-v1
Dataset Description
MSA-OmniVoice-v1 is a 50-hour synthetic Modern Standard Arabic (MSA) speech dataset generated using OmniVoice. The dataset contains high-quality synthetic speech from a single speaker paired with fully diacritized (تشكيل) transcripts.
It is intended for training and fine-tuning Arabic speech models, including Text-to-Speech (TTS), Automatic Speech Recognition (ASR), speech representation learning, and alignment tasks.… See the full description on the dataset page: https://huggingface.co/datasets/oddadmix/msa-omnivoice-tts-v1.homai_wake_word_omnivoice
Homai Wake Word OmniVoice
Synthetic two-label wake-word dataset generated with k2-fsa/OmniVoice using
cross-lingual voice cloning.
For every reference row from all train, validation, and test splits of:
bond005/sova_rudevices
bond005/sberdevices_golos_100h_farfield
the dataset contains two generated recordings:
Һомай, generated with OmniVoice language Bashkir;
Хомай, generated with OmniVoice language Russian.
Dataset structure
Split: train
Columns: audio… See the full description on the dataset page: https://huggingface.co/datasets/AigizK/homai_wake_word_omnivoice.omnivoice-zh
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
9,946
Total
19,946
omnivoice-it
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
10,000
Total
20,000
omnivoice-es
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
10,000
Total
20,000
omnivoice-vi
OmniVoice VI — Giọng Việt + SRT lồng tiếng
Dataset chứa 6 giọng tiếng Việt và công cụ speak.py để chạy trên Google Colab với OmniVoice.
Giọng có sẵn
Slug
Tên
ban_mai
Ban Mai
lan_trinh
Lan Trinh
ngan_ha
Ngan Ha
ngoc_huyen
Ngoc Huyen
thao_trinh
Thao Trinh
tuong_vy
Tuong Vy
Mỗi giọng gồm profile.json, voice.pt (prompt cache), audio mẫu và ref_text.txt.
Chạy trên Colab
Lồng tiếng SRT: mở colab/Omivoice_VI_Colab.ipynb
Clone… See the full description on the dataset page: https://huggingface.co/datasets/STBack23/omnivoice-vi.omnivoice-ja
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
9,958
Total
19,958
omnivoice-ru
Sample rate
24 kHz
Voice-designed
10,000
Voice-cloned
9,999
Total
19,999
omnivoice-audio-storageomnivoice-vi2
OmniVoice VI — Giọng Việt + SRT lồng tiếng
Dataset chứa 6 giọng tiếng Việt và công cụ speak.py để chạy trên Google Colab với OmniVoice.
Giọng có sẵn
Slug
Tên
ban_mai
Ban Mai
lan_trinh
Lan Trinh
ngan_ha
Ngan Ha
ngoc_huyen
Ngoc Huyen
thao_trinh
Thao Trinh
tuong_vy
Tuong Vy
Mỗi giọng gồm profile.json, voice.pt (prompt cache), audio mẫu và ref_text.txt.
Chạy trên Colab
Lồng tiếng SRT: mở colab/Omivoice_VI_Colab.ipynb
Clone… See the full description on the dataset page: https://huggingface.co/datasets/TruongVuVan/omnivoice-vi2.omnivoice-test-th
OmniVoice Test - Thai
Synthetic Thai speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz
omnivoice_gen_test
omnivoice_gen_test
Uzbek TTS evaluation set for OmniVoice voice cloning: 30 number-heavy prompts x
6 reference voices = 180 generations.
Every prompt deliberately mixes digits, times, dates, prices, phone numbers and
scores, so the set stresses number verbalisation as much as voice similarity.
Columns
column
type
description
id
string
ref{voice}_t{prompt}, e.g. ref3_t17
reference_id
string
reference voice, 1-6
text_id
int
prompt index, 0-29… See the full description on the dataset page: https://huggingface.co/datasets/openbank-uz/omnivoice_gen_test.omnivoice_deepfake_dataset
omnivoice_deepfake_dataset
210.4 hours of synthetic speech deepfakes — 179,868 clips across 6 corpora and 5 languages (Mandarin, Spanish, French, Italian, Japanese), generated via voice cloning and hundreds of designed voice profiles.
Generated with OmniVoice, built to train/evaluate speech deepfake detectors (part of the Forensics model family).
Built to improve detection of voice cloning, deepfakes, highly realistic synthetic voices, voice conversion/changing, and similar… See the full description on the dataset page: https://huggingface.co/datasets/eliya/omnivoice_deepfake_dataset.omnivoice-vie
OmniVoice VI — Giọng Việt + SRT lồng tiếng
Dataset chứa 6 giọng tiếng Việt và công cụ speak.py để chạy trên Google Colab với OmniVoice.
Giọng có sẵn
Slug
Tên
ban_mai
Ban Mai
lan_trinh
Lan Trinh
ngan_ha
Ngan Ha
ngoc_huyen
Ngoc Huyen
thao_trinh
Thao Trinh
tuong_vy
Tuong Vy
Mỗi giọng gồm profile.json, voice.pt (prompt cache), audio mẫu và ref_text.txt.
Chạy trên Colab
Mở notebook colab/Omivoice_VI_Colab.ipynb
Đặt HF_REPO =… See the full description on the dataset page: https://huggingface.co/datasets/hoanglinhn0/omnivoice-vie.omnivoice-voicebank
Mimba OmniVoice Voice Bank — Reusable Multi-Speaker Reference Voices
A bank of distinct synthetic reference voices, generated with
OmniVoice Voice Design (attribute-driven
generation, no reference audio). Each entry is a short reference clip + its
attributes + a speaker embedding.
The bank is language-agnostic: it captures only the timbre of each voice. It
is meant to be reused as ref_audio to generate multi-speaker TTS corpora in any
language (Plateau Malagasy plt, Ngiemboon… See the full description on the dataset page: https://huggingface.co/datasets/mimba/omnivoice-voicebank.EchoTTS-OmniVoice-En
EchoTTS + OmniVoice English
Synthetic English conversational speech dataset with 200 audio samples.
100 EchoTTS — unique random voices generated with EchoTTS (no speaker reference, seed-based voice diversity)
100 OmniVoice clones — voice clones using EchoTTS audio as reference, generated with OmniVoice
EchoTTS sample rate: 44.1 kHz
OmniVoice sample rate: 24 kHz
Language: English
EchoTTS-OmniVoice-en-20komni-voice-trainingspeechsynth-omnivoice-v2omnivoice-best-of-n-dev-eval
🎙 Best-of-N Voice Cloning Training Data
Curated training dataset for fine-tuning OmniVoice for IWSLT 2026.
🎧 Listen to the Audio
This dataset has playable audio columns — click on any row in the dataset viewer
to listen to both the reference audio (original speaker) and the best synthesized audio
(selected by quality score).
Dataset Description
For each sentence in the dev split of ymoslem/acl-6060 (884 samples × 3 languages),
we synthesized audio with the… See the full description on the dataset page: https://huggingface.co/datasets/amanuelbyte/omnivoice-best-of-n-dev-eval.omnivoice-test-zh
OmniVoice Test - Chinese
Synthetic Chinese speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz
omnivoice-test-tr
OmniVoice Test - Turkish
Synthetic Turkish speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz
omnivoice-test-ru
OmniVoice Test - Russian
Synthetic Russian speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz
omnivoice-test-de
OmniVoice Test - German
Synthetic German speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz
omnivoice-test-fr
OmniVoice Test - French
Synthetic French speech dataset generated with OmniVoice.
100 voice designs (unique synthetic speakers)
100 voice clones (cloned from voice design speakers)
Sample rate: 24 kHz