shangeth/jenny-mimi-codes

Jenny TTS — Kyutai Mimi Encoded

Jenny TTS pre-encoded with the Kyutai Mimi neural audio codec.

Instead of raw waveforms, every utterance is stored as a compact matrix of discrete codec tokens. This format is ready to use directly in any language-model-style audio generation pipeline without needing a GPU encoder at training time.

What's inside

manifest.jsonl       # metadata — one JSON record per utterance
shards/
├── shard_0000.pt    # packed dict of { idx -> (8, L) int16 code tensor }
├── shard_0001.pt
└── ...

Each manifest.jsonl record:

{
  "idx": 0,
  "text": "Welcome to the Jenny TTS dataset.",
  "codes_file": "shards/shard_0000.pt:0",
  "speaker_id": "jenny",
  "n_frames": 287
}

Dataset details

Source	reach-vb/jenny_tts_dataset
Speaker	Single female speaker (Jenny)
Total duration	~25 hours
Codec	Kyutai Mimi
Codec sample rate	24,000 Hz
Codec frame rate	12.5 fps
Codebooks	8
Token dtype	int16
License	See original dataset

What you can use this for

• Language-model-style TTS (autoregressive token prediction)
• Codec language model pre-training / fine-tuning
• Voice style transfer research
• Audio tokenization benchmarks
• Any task that benefits from a high-quality, single-speaker English speech corpus in discrete token form