boffire/f5tts-kabyle-dataset

F5-TTS Kabyle Dataset

Clean, deduplicated audio-text dataset for Kabyle (Taqbaylit / Tamaziɣt) TTS fine-tuning with F5-TTS.

Statistics

Metric	Value
Total clips	59,462
Total duration	41.30 hours
Sample rate	24 kHz mono
Avg clip length	2.50s
Min clip length	1.00s
Max clip length	12.65s
Unique phrases	59,462 (0% duplicates)
Unique characters	112
Sources	Tatoeba (67.8%) + Common Voice 26 tiny (32.2%)

Source Datasets

Dataset	Clips	Duration	License
boffire/tatoeba-kabyle-audio	40,293	21.13h	CC BY 4.0
boffire/common-voice-scripted-speech-kab-26-tiny	19,169	24.75h	Mixed

Preparation Pipeline

1. Resampled both sources to 24 kHz mono (F5-TTS / Vocos requirement)
2. Duration filter: 1.0s — 15s (removed 230 clips: 225 Tatoeba + 5 CV)
3. Character standardization: corrected false friends (ε→ɛ, γ→ɣ, etc.)
4. Deduplication: one instance per phrase, preferring CV over Tatoeba for speaker diversity
5. Balanced: ~68% Tatoeba / ~32% CV

File Format

metadata.csv (separator |)

audio_file|text
wavs/tatoeba_033368.wav|Uɣeɣ tannumi d uzɣal.
wavs/cv_024698.wav|Tanemmirt i Mass Nagata.
wavs/tatoeba_022648.wav|Tom iteṭṭef dima deg awal.

vocab.txt

Extended vocabulary merging F5-TTS base vocab + all Kabyle characters:

• ɛ Ɛ ɣ Ɣ č Č ǧ Ǧ
• ṭ Ṭ ḍ Ḍ ṛ Ṛ ẓ Ẓ
• ḥ Ḥ ṣ Ṣ ḏ Ḏ ṯ Ṯ

Usage

Load as audio dataset

from datasets import load_dataset

# Load with audio
ds = load_dataset("audiofolder", data_dir=".", split="train")

# Or load metadata only
df = pd.read_csv("metadata.csv", sep="|")

For F5-TTS fine-tuning

# Prepare F5-TTS format
python prepare_csv_wavs.py ./f5tts_kabyle_dataset ./data/kabyle_char

# Extend checkpoint
python extend_checkpoint.py \
    --checkpoint ckpts/F5TTS_Base/model_1200000.safetensors \
    --old_vocab ckpts/F5TTS_Base/vocab.txt \
    --new_vocab ./data/kabyle_char/vocab.txt \
    --output ckpts/F5TTS_Base/model_1200000_extended.safetensors

# Fine-tune
accelerate launch --mixed_precision=fp16 \
    src/f5_tts/train/train.py \
    --config-name F5TTS_Kabyle.yaml

Character Coverage

All standard Kabyle characters are present:

Char	Unicode	Frequency
ɣ	U+0263	~35,941
ḍ	U+1E0D	~10,141
ɛ	U+025B	present
č	U+010D	present
ǧ	U+01E7	present
ṭ	U+1E6D	present
ṛ	U+1E5B	present
ẓ	U+1E93	present

Dataset Quality

• Zero duplicates (59,462 unique phrases)
• Multi-speaker (CV provides voice diversity)
• Clean audio (Tatoeba TTS + CV scripted speech)
• Verified alignment (5 random samples manually checked)
• Standardized orthography (no Greek/Cyrillic false friends)

Limitations

• Tatoeba is single-speaker (synthetic/reading voice)
• CV-tiny is a subset of full Common Voice 26 Kabyle
• Average clip length is short (2.50s) — good for alignment, less for prosody
• No speaker labels in metadata (for multi-speaker conditioning)

Future Work

• Add speaker embeddings for multi-speaker TTS
• Integrate full CV26 dataset (~23h additional)
• Add phonetic transcription (IPA) for better pronunciation
• Create train/dev/test splits with speaker disjointness

Citation

If you use this dataset, please cite:

@dataset{f5tts_kabyle_2026,
  author = {MOKRAOUI, Athmane (boffire)},
  title = {F5-TTS Kabyle Dataset},
  year = {2026},
  publisher = {HuggingFace},
  url = {https://huggingface.co/datasets/boffire/f5tts-kabyle-dataset}
}

And the source datasets:

• Tatoeba Project
• Mozilla Common Voice

License

• Tatoeba portion: CC BY 4.0
• Common Voice portion: Mixed (see original dataset licenses per clip)
• This compilation: CC BY 4.0

Contact

• Author: Athmane MOKRAOUI (boffire / ButterflyOfFire)
• GitHub: athmanemokraoui
• HuggingFace: boffire

---

Prepared for F5-TTS fine-tuning on Google Colab T4 / Kaggle T4 (16 GB VRAM)