feat(speech): add pocket-tts CPU text-to-speech pipeline

[sandyz1000]

Summary

Adds pocket-tts as a second speech pipeline alongside Dia, over the same OpenAI /v1/audio/speech contract. It's a Kyutai-style Mimi codec + FlowLM transformer that runs fast on CPU, so one mistral.rs binary can serve an LLM plus a CPU-fast TTS (Dia is GPU-oriented).

Selectable via serve -m (auto-detected), serve speech, or from-config; voice defaults to alba and is settable via --voice / config / SDK. /v1/audio/speech now routes by the request's model, so pocket-tts (CPU) and Dia (GPU) can be loaded together and picked per request.

Attribution

Model + inference code under mistralrs-core/src/speech_models/pockettts/ is ported from the pocket-tts Rust crate: https://github.com/babybirdprd/pocket-tts — mostly near-verbatim, recompiled against this repo's candle 0.11 (upstream pins 0.9.1). The rest is the mistral.rs integration layer.

Source is MIT OR Apache-2.0, this repo is MIT — compatible. Attribution is also in the module header. Opening transparently so maintainers (and the pocket-tts author, if they wish) can review provenance.

Verification

• cargo check clean (0 warnings) and cargo fmt --all --check pass.
• End-to-end on Apple Metal and over HTTP: kyutai/pocket-tts-without-voice-cloning produces healthy 24 kHz mono speech.

[github-actions]

Code Metrics Report

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 Language              Files        Lines         Code     Comments       Blanks
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 C Header                 46         9600         7138         1327         1135
 C++                       1           36           30            1            5
 C++ Header                6         4022         3237          399          386
 CSS                       5         1069          918            6          145
 CUDA                    143        38519        32957         1775         3787
 Dockerfile                1           35           19            9            7
 HTML                      3          125          119            0            6
 JavaScript                8         1241         1186           11           44
 Jinja2                    7          694          656            5           33
 JSON                     34        26374        26371            0            3
 Makefile                  1           18           16            0            2
 MDX                      37         6975            0         5157         1818
 Metal Shading Lan|       37        14422        11414         1136         1872
 PowerShell                1          657          571           31           55
 Python                  161        34938        31448          487         3003
 Shell                     3         1071          852          115          104
 Plain Text               54        10714            0         9231         1483
 TOML                     28         1399         1212           43          144
 TypeScript               11         1658         1418           66          174
 YAML                      3           25           23            2            0
─────────────────────────────────────────────────────────────────────────────────
 Jupyter Notebooks         3          122           83           23           16
 |- Markdown               1           60           30           22            8
 |- Python                 1          122          113            1            8
 (Total)                              304          226           46           32
─────────────────────────────────────────────────────────────────────────────────
 Markdown                284        13126            0         9834         3292
 |- BASH                  27          349          268           48           33
 |- Dockerfile             2           14           12            0            2
 |- JSON                   6          292          292            0            0
 |- PowerShell             1            1            1            0            0
 |- Python               135         7349         6119          306          924
 |- Rust                  63         3873         2873          398          602
 |- TOML                   7          116           97            0           19
 (Total)                            25120         9662        10586         4872
─────────────────────────────────────────────────────────────────────────────────
 Rust                    774       501441       457490         5598        38353
 |- Markdown             445        10408          452         8767         1189
 (Total)                           511849       457942        14365        39542
─────────────────────────────────────────────────────────────────────────────────
 Svelte                   19         1974         1832           50           92
 |- CSS                    1            4            4            0            0
 |- JavaScript            19          928          773           25          130
 (Total)                             2906         2609           75          222
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 Total                  1670       693771       590024        44873        58874
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━