--- license: cc-by-nc-4.0 gated: manual extra_gated_heading: Acknowledge the license to access this dataset extra_gated_prompt: This dataset is released under CC-BY-NC-4.0 — free for personal and research use. Commercial use is prohibited under this license; for a commercial license, contact eliya@vocos.io. extra_gated_description: Requests are reviewed manually. extra_gated_button_content: Agree and request access extra_gated_fields: I agree to the non-commercial license terms above and understand commercial use requires a separate license: checkbox Intended use (one sentence): text task_categories: - audio-classification tags: - audio - deepfake - audio-deepfake-detection - anti-spoofing - voice - text-to-speech - qwen3 pretty_name: Qwen3-TTS Speech Deepfake Dataset dataset_info: - config_name: Aishell features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 6685037543.494 num_examples: 45738 - name: test num_bytes: 355194148.552 num_examples: 2424 download_size: 6801748267 dataset_size: 7040231692.0460005 - config_name: Aishell3 features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 4396562105.464 num_examples: 33788 - name: test num_bytes: 238054114.4 num_examples: 1834 download_size: 4449557872 dataset_size: 4634616219.863999 - config_name: CommonVoice features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 12283256142.584 num_examples: 83542 - name: test num_bytes: 614216694.174 num_examples: 4138 download_size: 12561736109 dataset_size: 12897472836.758 - config_name: LibriTTS features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 15201952566.228 num_examples: 50358 - name: test num_bytes: 782720188.74 num_examples: 2630 download_size: 15683556803 dataset_size: 15984672754.968 - config_name: VCTK-Corpus features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 5572158278.87 num_examples: 44738 - name: test num_bytes: 310179649.552 num_examples: 2534 download_size: 5743491022 dataset_size: 5882337928.422 - config_name: commonvoice_es features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 2876680987.18 num_examples: 22370 - name: test num_bytes: 148874729.886 num_examples: 1086 download_size: 2900862696 dataset_size: 3025555717.066 - config_name: commonvoice_fr features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 2641509456.91 num_examples: 21730 - name: test num_bytes: 137069121.756 num_examples: 1138 download_size: 2623777343 dataset_size: 2778578578.666 - config_name: commonvoice_it features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 3374819591.312 num_examples: 22004 - name: test num_bytes: 172206276.144 num_examples: 1164 download_size: 3416374861 dataset_size: 3547025867.456 - config_name: commonvoice_ja features: - name: audio dtype: audio - name: duration dtype: float32 - name: method dtype: string - name: label dtype: string - name: dataset dtype: string - name: model dtype: string - name: sample_id dtype: string - name: predicted_age dtype: int32 - name: predicted_gender dtype: string - name: speaker dtype: string - name: orig_file dtype: string - name: donor_file dtype: string splits: - name: train num_bytes: 3500721773.624 num_examples: 22128 - name: test num_bytes: 202026062.136 num_examples: 1296 download_size: 3466892532 dataset_size: 3702747835.76 configs: - config_name: Aishell data_files: - split: train path: Aishell/train-* - split: test path: Aishell/test-* - config_name: Aishell3 data_files: - split: train path: Aishell3/train-* - split: test path: Aishell3/test-* - config_name: CommonVoice data_files: - split: train path: CommonVoice/train-* - split: test path: CommonVoice/test-* - config_name: LibriTTS data_files: - split: train path: LibriTTS/train-* - split: test path: LibriTTS/test-* - config_name: VCTK-Corpus data_files: - split: train path: VCTK-Corpus/train-* - split: test path: VCTK-Corpus/test-* - config_name: commonvoice_es data_files: - split: train path: commonvoice_es/train-* - split: test path: commonvoice_es/test-* - config_name: commonvoice_fr data_files: - split: train path: commonvoice_fr/train-* - split: test path: commonvoice_fr/test-* - config_name: commonvoice_it data_files: - split: train path: commonvoice_it/train-* - split: test path: commonvoice_it/test-* - config_name: commonvoice_ja data_files: - split: train path: commonvoice_ja/train-* - split: test path: commonvoice_ja/test-* --- # qwen3_deepfake_dataset **515.5 hours of synthetic speech deepfakes** — 364,640 clips across 9 corpora and 6 languages (Mandarin, English, Spanish, French, Italian, Japanese), generated via voice cloning and hundreds of designed voice profiles. Generated with **Qwen3-TTS**, built to train/evaluate speech deepfake detectors (part of the [Forensics model family](https://huggingface.co/collections/eliya/forensics-speech-deepfake-detection-family)). Built to improve detection of voice cloning, deepfakes, highly realistic synthetic voices, voice conversion/changing, and similar identity-manipulation techniques. ## How it was built Real utterances from 9 public speech corpora (Aishell, Aishell3, CommonVoice — en/es/fr/it/ja, LibriTTS, VCTK-Corpus) were fed through Qwen3-TTS in three modes: - **clone_self** — resynthesize a speaker's own line using their own voice as the reference - **clone_cross** — resynthesize a speaker's line using a *different* donor speaker's voice as the reference (identity-swap deepfake) - **design** — synthesize speech from a designed target voice profile (age/gender/pitch), not cloned from any specific reference Every generated clip's speaker age and gender were predicted and attached as columns. The metadata additionally includes one `bonafide` row per unique real source utterance, for reference — **the audio for those rows is not included in this repo** (see below). ## Example: one source recording, every variant generated from it Since this dataset is gated, here's exactly what you're requesting access to — a real example, not a made-up one. One Aishell recording (`BAC009S0002W0135`) and everything generated from it: | File | Method | What it is | |---|---|---| | `speechfake/Real/Aishell/train/S0002/BAC009S0002W0135.wav` | *(original — not bundled, see above)* | The real recording everything below is derived from | | `Aishell/BAC009S0002W0135/BAC009S0002W0135_clone_self.wav` | clone_self | Same content, resynthesized in the *same* speaker's own voice | | `Aishell/BAC009S0002W0135/BAC009S0002W0135_clone_cross.wav` | clone_cross | Same content, resynthesized in a *different* donor speaker's voice — an identity-swap deepfake | | `..._designed_f_elder70s_lowpitched_v_379.wav` | design | Same content, synthesized from an invented voice profile (female, elderly, low-pitched) — no real reference voice at all | | `..._designed_m_early60s_mediumlow_vo_327.wav` | design | Same content, a different invented profile (male, early 60s, medium-low pitch) | | `..._designed_m_late60s_medium_voice_012.wav` | design | Another invented profile (male, late 60s, medium pitch) | | `..._designed_m_mid50s_medium_voice_057.wav` | design | Another invented profile (male, mid-50s, medium pitch) | `design` draws from hundreds of predefined voice-profile descriptions (age bracket × gender × pitch register) rather than cloning any real person — the same source line gets resynthesized in dozens of different invented voices, as shown above. ## What's in this repo - The **generated fake audio** (`.wav`), organized as `//.wav`. - `metadata/metadata_age_gender_train.csv` and `metadata/metadata_age_gender_test.csv` — one row per clip. ## What's *not* in this repo Rows labeled `bonafide` describe the original real recordings this dataset was built from. Their `new_file`/`orig_file`/`donor_file` columns point at paths like `speechfake/Real/Aishell/train/S0002/BAC009S0002W0135.wav` — that audio lives in its own original public dataset (Aishell, Aishell3, CommonVoice, LibriTTS, or VCTK-Corpus, matching the `dataset` column) and is not duplicated here. Fetch it from the original source if you need the real counterpart. ## Columns | column | meaning | |---|---| | `new_file` / `file` | path to this row's audio (relative to repo root for `fake` rows; a reference-only path for `bonafide` rows — see above) | | `orig_file` | the real utterance this clip was cloned from (reference-only path) | | `donor_file` | for `clone_cross`, the donor speaker's reference utterance (reference-only path) | | `duration` | seconds | | `method` | `qwen_clone_self`, `qwen_clone_cross`, `qwen_design`, or `real` | | `label` | `fake` or `bonafide` | | `dataset` | source corpus | | `predicted_age` / `predicted_gender` | model-predicted speaker attributes | | `speaker` / `sample_id` | source speaker/utterance identifier | ## Use This dataset is gated — request access on this page, then authenticate before downloading: ```bash huggingface-cli login # or: export HF_TOKEN=hf_xxx ``` Load the audio directly (organized as 9 configs, one per source corpus): ```python from datasets import load_dataset, get_dataset_config_names get_dataset_config_names("eliya/qwen3_deepfake_dataset") # list the 9 corpora ds = load_dataset("eliya/qwen3_deepfake_dataset", "Aishell", split="train") sample = ds[0] sample["audio"]["array"], sample["audio"]["sampling_rate"] # decoded waveform + rate # For the larger corpora, stream instead of downloading everything upfront: ds = load_dataset("eliya/qwen3_deepfake_dataset", "CommonVoice", split="train", streaming=True) next(iter(ds)) ``` Or work from the full reference metadata directly (all 9 corpora, includes `bonafide` provenance rows): ```python import pandas as pd df = pd.read_csv("metadata/metadata_age_gender_train.csv") fake_only = df[df["label"] == "fake"] # bonafide rows have no bundled audio ``` ## License CC-BY-NC-4.0 — free for personal and research use. For commercial use, contact eliya@vocos.io.