Upload README.md with huggingface_hub

README.md

@@ -4,44 +4,170 @@ language: [eu]
 task_categories: [text-to-speech, automatic-speech-recognition]
 pretty_name: BSSD (Basque Synthetic Speech Dataset)
 size_categories: [100K<n<1M]
+tags: [synthetic, speech, basque, euskara, tts, omnivoice]
+dataset_info:
+  features:
+  - name: id
+    dtype: string
+  - name: audio
+    dtype:
+      audio:
+        sampling_rate: 24000
+  - name: text
+    dtype: string
+  - name: text_norm
+    dtype: string
+  - name: text_src
+    dtype: string
+  - name: voice_mode
+    dtype: string
+  - name: voice_id
+    dtype: string
+  - name: prompt_src
+    dtype: string
+  - name: instruct
+    dtype: string
+  - name: duration
+    dtype: float64
+  - name: cer_whisper
+    dtype: float64
+  - name: cer_conformer
+    dtype: float64
+  - name: utmos
+    dtype: float64
+  - name: spk_sim
+    dtype: float64
+  - name: teacher
+    dtype: string
+  - name: num_step
+    dtype: int64
+  - name: guidance
+    dtype: float64
+  - name: status
+    dtype: string
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*
 ---
 
 # BSSD: Basque Synthetic Speech Dataset
 
-**287.6 hours**, 222637 utterances, 11710 voice identities, 24 kHz mono FLAC.
+**364,870 utterances, 470.8 hours** of synthetic Basque (euskara) read speech, 24 kHz mono FLAC,
+with normalised transcripts, voice metadata and per-utterance quality scores. Version 1.0-rc,
+2026-10-17. Built by [itzune](https://huggingface.co/itzune) as training data for lightweight
+Basque TTS (first target: a [Pocket TTS](https://github.com/kyutai-labs/pocket-tts) student that
+runs in real time on CPU); also usable for ASR data augmentation and TTS research.
 
-## Provenance
+| | |
+|---|---|
+| utterances / hours | 364,870 / 470.8 h (mean 4.65 s) |
+| voice identities | ~18,400 ids, of which 180 have ≥ 1 h; 134 stable synthetic bank voices; 54 real speakers; 8 named default voices |
+| intelligibility (ASR round trip) | CER median 0.0, p90 0.013, p99 0.027 on both Basque ASRs |
+| naturalness proxy | UTMOS p10 3.16 / p50 3.51 / p90 3.90 |
+| text | 236 h Wikipedia, 130 h Common Voice sentences, 75 h news (EusCrawl, CC BY-SA sites), 24 h generated numbers/dates, 4.5 h Wikisource, 0.5 h Tatoeba |
 
-All audio was generated with [OmniVoice](https://github.com/k2-fsa/OmniVoice)
-(`k2-fsa/OmniVoice`, code Apache-2.0, weights CC-BY-NC, Higgs Audio V2 tokenizer under the Boson
-community license) at num_step=32, guidance_scale=2.0. Prompt modes: synthetic
-voice bank (OmniVoice auto/design clips used as cloning prompts), real Basque prompts from
-OpenSLR 76 (CC-BY-SA-4.0) and the HiTZ-Aholab studio corpus (CC-BY-4.0), voice design and
-auto voice. No Common Voice contributor's voice was cloned.
+## How it was made
 
-Text sources: Basque Wikipedia (CC-BY-SA-4.0), Mozilla Common Voice sentence pool (CC0),
-Basque Parliament transcripts (CC0), OpenSLR 76 transcripts (CC-BY-SA-4.0) and templated
-number/date sentences. Numbers and abbreviations were verbalised (`text_norm`).
+**Teacher.** All audio was generated with [OmniVoice](https://github.com/k2-fsa/OmniVoice)
+(`k2-fsa/OmniVoice`, commit `c5fdb5cc`, package built from git on 2026-10-15; 32 decoding steps,
+guidance 2.0, fp16) on two NVIDIA L40s. OmniVoice's code is Apache-2.0; its **weights are CC-BY-NC**
+(the authors cite training-data constraints such as Emilia) and it bundles Boson's Higgs Audio V2
+tokenizer under the Boson community license. We state this provenance so that downstream users can
+make their own assessment; see "License" below.
 
-## Filtering
+**Voices (`voice_mode`).**
+- `clone_bank` (213 h): 134 stable synthetic voices. OmniVoice generated 4,000 candidate voices in
+  its auto and voice-design modes; candidates were kept only with CER ≤ 2 %, UTMOS in the top half,
+  no clipping, then de-duplicated with WavLM x-vectors (cosine < 0.96, calibrated on real speakers) and
+  balanced by pitch. Each kept clip is then used as the cloning prompt, so every bank voice is
+  consistent across hundreds of sentences. Prompts are published in `voices.jsonl`.
+- `clone_named` (77 h): eight default voices with a reserved share. **Maider** (16.4 h) and
+  **Antton** (14.5 h) are the professional speakers of the
+  [HiTZ-Aholab studio corpus](https://zenodo.org/records/17952596) (CC BY 4.0), cloned from a 5–7 s
+  studio clip; **Nahia, Ane, Mikel, Jon, Unai, Aitor** (7–8 h each) are synthetic voices chosen by
+  listening among 84 voice-design candidates. The named voices were given extra hours on purpose so
+  that a student model reproduces them reliably as presets.
+- `clone_real` (126 h): clones of 54 real speakers who recorded for TTS: the 52 volunteers of
+  [OpenSLR 76](https://www.openslr.org/76/) (CC BY-SA 4.0) and the two HiTZ-Aholab speakers.
+  **No Common Voice contributor was cloned.**
+- `design` (33 h) and `auto` (22 h): one-off voices from OmniVoice's attribute-controlled and random
+  voice modes.
 
-Every clip was transcribed with `xezpeleta/whisper-large-v3-eu` (and, where present,
-`HiTZ/stt_eu_conformer_transducer_large_v2`); kept rows have CER <= 0.03 on all ASRs,
-hypothesis length ratio within [0.7, 1.3], UTMOS >= 3.0, speaker similarity to the
-prompt >= 0.9 (cloned rows), no clipping and no internal silence > 1.5s.
+**Text (`text_src`).** Sentences were filtered (4–30 words, Basque alphabet, a vocabulary check
+against 380k trusted transcripts for web text, no abbreviations or foreign-letter tokens, no
+duplicates) and normalised by a rule-based Basque front-end (`text_norm`: cardinals and ordinals
+following Euskaltzaindia, case suffixes after digits such as *1990ean → mila bederatziehun eta
+laurogeita hamarrean*, percentages, decimals, times, units, acronyms). Sentences with constructs the
+normaliser refuses to guess (ranges, numeric dates, fractions) were dropped. Sources: Basque
+Wikipedia (CC BY-SA), the Mozilla Common Voice Basque sentence pool (CC0), news and local media from
+the [EusCrawl](https://ixa.ehu.eus/euscrawl/) free release restricted to documents the publishers
+licensed CC BY-SA / CC BY (Berria, Goiena, Argia, Ataria, Guaixe, …), Basque Wikisource (CC BY-SA),
+Tatoeba (CC BY) and template-generated number/date sentences. Basque Parliament transcripts were
+**excluded** (lower-cased ASR fragments without punctuation).
+
+**Filtering (`status = keep` for every published row).** Each clip was transcribed with
+[`xezpeleta/whisper-large-v3-eu`](https://huggingface.co/xezpeleta/whisper-large-v3-eu) (CT2) and
+[`HiTZ/stt_eu_conformer_transducer_large_v2`](https://huggingface.co/HiTZ/stt_eu_conformer_transducer_large_v2);
+rows were kept with CER ≤ 3 % on both, hypothesis length ratio 0.7–1.3, 8–22 characters per second,
+UTMOS ≥ 3.0, speaker similarity to the prompt ≥ 0.90 (cloned rows), no clipping, no internal
+silence > 1.5 s, duration 1.5–20 s. About 85 % of generated rows passed; a per-voice cap of 1.5 h
+was applied to non-named voices; two rows were removed after listening (`alegia` mispronounced,
+one garbled utterance). Generated rows that failed are not included.
+
+**Does OmniVoice reproduce real Common Voice speakers?** Common Voice (CC0) is in OmniVoice's
+training data. We compared every auto and design voice (4,000 clips) with one clip from each of
+the 9,678 Basque Common Voice v20 contributors using ECAPA-TDNN speaker embeddings. The nearest
+contributor to any random voice scored at most 0.78 (auto) while the 99th percentile of nearest
+*different* real contributors is 0.805; none of the 2,400 auto voices and 1 of 1,600 design voices
+(0.06 %, below chance) exceeded that line. The random voices are statistically new speakers.
 
 ## Columns
 
-id, audio, text, text_norm, text_src, voice_mode, voice_id, prompt_src, instruct, duration,
-cer_whisper, cer_conformer, utmos, spk_sim, teacher, num_step, guidance, status.
+| column | meaning |
+|---|---|
+| `id` | `eu_syn<run>_<shard>_<n>` |
+| `audio` | 24 kHz mono FLAC, trimmed, peak-normalised |
+| `text` / `text_norm` | sentence as written / as sent to the teacher (verbalised numbers etc.) |
+| `text_src` | `euwiki`, `cv`, `euscrawl`, `gen_numbers`, `wikisource`, `tatoeba` |
+| `voice_mode`, `voice_id`, `prompt_src`, `instruct` | see above; `prompt_src` is `bank:*`, `named:*`, `oslr76:*`, `hitz:*` |
+| `duration` | seconds |
+| `cer_whisper`, `cer_conformer` | ASR round-trip character error rate vs `text_norm` |
+| `utmos`, `spk_sim` | predicted MOS; WavLM cosine similarity to the prompt (cloned rows) |
+| `teacher`, `num_step`, `guidance`, `status` | generation parameters; `keep` |
+
+Extra files: `voices.jsonl` (bank voices and prompts), `pocket_tts_manifest-train-vm1.jsonl` /
+`-vm2.jsonl` (path, duration, transcript manifests used for Pocket TTS training).
 
-## Limitations
+## Known limitations
 
-Synthetic read speech: prosody and voice variety are those of the teacher; no spontaneous
-speech; occasional teacher artifacts may remain. Intended for TTS/ASR research and training.
+- Synthetic read speech: prosody and timbre variety are those of the teacher; no spontaneous speech.
+- Occasional Spanish-style readings of Basque orthography survive ASR filtering (e.g. *ge/gi* read as
+  /xe, xi/ in ~0.1 % of such rows; the name *Julen* read with a fricative). Numbers are verbalised by
+  rules and may differ from a speaker's choice in rare cases.
+- Mean duration is short (4.65 s); long-form prosody is under-represented.
+- The non-named voices come from a model whose Basque training audio is mostly crowd-sourced, so
+  their acoustic quality is "clean crowd microphone" rather than studio, except for Maider, Antton
+  and the real-speaker clones.
 
 ## License and attribution
 
-CC-BY-SA-4.0. Please credit: k2-fsa OmniVoice; Wikimedia / Basque Wikipedia editors;
-Mozilla Common Voice; Google (OpenSLR 76, Kjartansson et al. 2020); HiTZ Zentroa / Aholab;
-Basque Parliament corpus (UPV/EHU GTTS).
+The dataset is released under **CC BY-SA 4.0**. Please credit: *BSSD (Basque Synthetic Speech
+Dataset), itzune, 2026*, and the sources it builds on: k2-fsa OmniVoice (Zhu et al., 2026,
+arXiv:2604.00688); Wikimedia (Basque Wikipedia and Wikisource); Mozilla Common Voice; the EusCrawl
+corpus (Artetxe et al., 2022) and the publishers Berria, Goiena, Argia and others; Google /
+OpenSLR 76 (Kjartansson et al., 2020); HiTZ Zentroa – Aholab (studio corpus, speakers Maider and
+Antton); Tatoeba. Users should be aware that the teacher model's weights are licensed CC-BY-NC by
+their authors; we document the full chain above rather than claim a cleaner one.
+
+## Citation
+
+```
+@misc{bssd2026,
+  title  = {BSSD: Basque Synthetic Speech Dataset},
+  author = {itzune},
+  year   = {2026},
+  url    = {https://huggingface.co/datasets/itzune/bssd}
+}
+```