DualCodec — Turkish, 25 Hz
DualCodec neural speech codec fine-tuned on Turkish, operating at 25 Hz.
The Turkish half of the pair with
dualcodec-hindi-25hz.
Part of the TinyAya codec study into whether language-specific codec fine-tuning improves reconstruction for low-resource languages — directly relevant to the S2ST model, whose audio fidelity is limited by its frozen depth decoder.
Produced by codec-finetuning. Upstream DualCodec
licence terms apply.
Code
| repo | what it does |
|---|---|
codec-finetuning |
fine-tunes Mimi / DualCodec / Kanade on Turkish + Hindi |
Project
TinyAya Stage 2 — Turkish⇄Hindi speech-to-speech translation with a text inner-monologue: a LoRA-adapted Cohere2 backbone driving a frozen Moshi depth decoder over Mimi codes.
The v0.3 run covered 76,250 steps / 2.07 epochs on a Cloud TPU v6e-16 (best val composite 2.8199 @ step 76,000). Read honestly: the text inner-monologue learns to translate (free-run chrF++ ~25.7 / 25.1), while intelligible audio synthesis remains the frontier (ASR-chrF++ 3.7 / 9.6 against a 92.1 / 86.6 ground-truth-audio ceiling) — bounded by the frozen depth decoder, not by translation understanding.
- Results: v0.3 evaluation report
- Training run: W&B
xzcb60bl· emergence report - Blog: Adapting Moshi for Low-Resource Speech Translation
Compute for the v0.3 run was provided by Google's TPU Research Cloud (TRC).