Soundbox

Kyutai · Voice

Pocket TTS — English

Commercial use: conditions applyCC-BY-4.0 (weights) + MIT (code)

The only entry in the catalog with a named voice list (26 voices), and the only one with two layers of licence. The weights are CC-BY-4.0 and the code is MIT. The voices are licensed separately, per collection: Kyutai's voice repository ranges from CC0 through CC-BY-4.0 to CC-BY-NC-4.0, so the speaker you pick can decide whether your use is commercial-safe.

kyutai/pocket-tts on Hugging Faceaudio.cpp

The licence

Summarized from the linked source documents, last checked 25 August 2026. Every claim links to its source; anything unconfirmed is marked unconfirmed.

CC-BY-4.0 (weights) + MIT (code)

Commercial use: conditions apply

The weights are CC-BY-4.0 and the code is MIT, so commercial use is permitted with attribution. Two things complicate it: Kyutai's original repo is gated behind prohibited-use terms (Soundbox downloads an ungated GGUF conversion, so you do not see that prompt), and the bundled speaker voices carry their own separate licences.

Conditions

  • Attribution, per CC-BY-4.0.
  • The upstream Kyutai gate prompt prohibits "voice impersonation or cloning without explicit and lawful consent", and misinformation or deception including presenting generated audio as a genuine recording. Kyutai disclaims all liability for non-compliant use.
  • The voices are licensed separately from the model. Kyutai's kyutai/tts-voices repo is per-collection: voice-zero and voice-donations are CC0, vctk and cml-tts are CC-BY-4.0, but expresso and ears are CC-BY-NC-4.0 (non-commercial). Which collection a given speaker id comes from decides whether your use is commercial-safe.

What about the audio you generate?

No output-ownership clause and no watermarking statement found on the model card. No output terms could be confirmed either way.

Correction to the shipped catalog: The English package's speaker ids (alba, cosette, eponine, fantine, javert, marius, peter_yearsley and the rest) look like public-domain LibriVox readers and Les Misérables characters, which would put them in the CC0 voice-zero collection. The per-name mapping could not be confirmed: the voice-zero collection is not a fetchable Hugging Face repo, and there is a separate CC-BY-4.0 "alba-mackenna" collection that may cover at least one of those names. Treat the individual voice licences as unconfirmed.

Based on the licence text as of the date below. Licences change; the links carry the authoritative version.

Specs

Transcribed from the shipped catalog, audiocpp_models_catalog.json — that file is authoritative if a number here disagrees with it.

Vendor
Kyutai
Catalog id
audiocpp-pocket-tts
audio.cpp loader family
pocket_tts
Can do
Text to speech
Weights repo
audio-cpp/audio.cpp-gguf
Weights (a package)
PocketTTS-GGUF/english/pocket-tts-english-q8_0.gguf, PocketTTS-GGUF/english/embeddings/**
Account needed to download
No
Max clip length
60s
Sample rate
44.1kHz mono
Approx. VRAM
2.0 GB
In the picker by default
No — opt in from Settings
Run end-to-end here
No — catalogued, not tested here

Getting it

There is no per-model install. One native binary runs all of these, so getting a model is a download of its weights and nothing else — no virtualenv, no torch pin, no CUDA wheel to match against your driver.

From the CLI
soundbox model install audiocpp
soundbox pull audiocpp-pocket-tts
  • The binary is fetched from audio.cpp's GitHub releases the first time you ask for it and cached under your own data directory. Soundbox ships no native binaries and vendors none.
  • Weights land in <models dir>/audiocpp/audiocpp-pocket-tts/ rather than in the shared Hugging Face cache, so a download stays scoped to the model you asked for.
  • This one is opt-in: switch it on in the Models page's “Enable more models” section before it appears in the picker.

Worth knowing

  • The entry pulls a speaker-embedding directory alongside the GGUF, which is why its weights glob is a list.
  • Opt-in and untested here.

Sources

Primary sources: the model's own repository, the project it came from, and the conversion audio.cpp actually downloads. Never invented — an entry with nothing findable says so instead.

Licence details on these pages are summaries of the linked source documents, last checked 25 August 2026. They are not legal advice, and Soundbox is not affiliated with any model's authors — weights download from Hugging Face under the user's own account and terms. Company and model names identify whose work it is, and belong to their owners. The linked licence text is authoritative; commercial use should be checked against it, per model, before anything ships.

Related models

Same loader family (pocket_tts)

Other voice models