BricksDisplay/OuteTTS-0.2-500M-GGUF

OuteTTS-0.2-500M — GGUF

OuteAI's OuteTTS-0.2-500M with its reference voices built in.

Two files make one model, as with llama.cpp's multimodal projectors: the model file is the text-to-speech LM backbone, run by llama.cpp; the codec file holds the audio side (decoder / vocoder, speaker encoders and the adaptor between the two), run by codec.cpp. The runtimes below wire the two graphs together, on CPU or GPU (CUDA, Vulkan, Metal), fully offline.

Files

Every run needs one model file + one codec file from this repo.

	--backbone	--model (codec)	Total
Recommended	OuteTTS-0.2-500M-Q8_0.gguf	codec-q8_0.gguf	672 MB
Smallest	OuteTTS-0.2-500M-Q4_K_M.gguf	codec-q4_k_m.gguf	515 MB
Best quality	OuteTTS-0.2-500M-FP16.gguf	codec-f16.gguf	1.2 GB

The model files are OuteAI's own GGUFs, unchanged. The codec file carries this model's voices; use it only with this model.

Other files

File	Size
OuteTTS-0.2-500M-Q2_K.gguf	344 MB
OuteTTS-0.2-500M-Q3_K_L.gguf	375 MB
OuteTTS-0.2-500M-Q3_K_M.gguf	361 MB
OuteTTS-0.2-500M-Q3_K_S.gguf	344 MB
OuteTTS-0.2-500M-Q4_0.gguf	358 MB
OuteTTS-0.2-500M-Q4_1.gguf	380 MB
OuteTTS-0.2-500M-Q4_K_S.gguf	391 MB
OuteTTS-0.2-500M-Q5_0.gguf	402 MB
OuteTTS-0.2-500M-Q5_1.gguf	425 MB
OuteTTS-0.2-500M-Q5_K_M.gguf	426 MB
OuteTTS-0.2-500M-Q5_K_S.gguf	418 MB
OuteTTS-0.2-500M-Q6_K.gguf	511 MB
codec-f32.gguf	339 MB
codec-q5_k_m.gguf	118 MB

Command line (codec.cpp)

With codec.cpp's tts-cli (build it; it bundles llama.cpp), then:

hf download BricksDisplay/OuteTTS-0.2-500M-GGUF OuteTTS-0.2-500M-Q8_0.gguf codec-q8_0.gguf --local-dir OuteTTS-0.2-500M

./build/tts-cli synthesize --backbone OuteTTS-0.2-500M/OuteTTS-0.2-500M-Q8_0.gguf --model OuteTTS-0.2-500M/codec-q8_0.gguf \
    --text "The birch canoe slid on the smooth planks." --output out.wav

List and pick voices:

./build/tts-cli voices --model OuteTTS-0.2-500M/codec-q8_0.gguf
./build/tts-cli synthesize ... --voice NAME --text "..." --output out.wav

All tts-cli options (voices, voice cloning, sampling, GPU) are described in the codec.cpp README.

Python

The codec_cpp package runs the same pipeline (it builds codec.cpp with its bundled llama.cpp, so it needs CMake and a C++ compiler):

pip install "codec-cpp @ git+https://github.com/mybigday/codec.cpp" huggingface_hub soundfile

import codec_cpp, soundfile
from huggingface_hub import hf_hub_download

tts = codec_cpp.TTS(hf_hub_download("BricksDisplay/OuteTTS-0.2-500M-GGUF", "OuteTTS-0.2-500M-Q8_0.gguf"),
                    hf_hub_download("BricksDisplay/OuteTTS-0.2-500M-GGUF", "codec-q8_0.gguf"))
audio, sr = tts.synthesize("The birch canoe slid on the smooth planks.")
soundfile.write("out.wav", audio, sr)

tts.voices() lists the builtin voices; ref_audio= clones a voice where the model supports it; tts.stream(text) yields audio as it is generated. More in the Python docs.

React Native (llama.rn)

llama.rn loads the same two files on iOS and Android:

const ctx = await initLlama({ model: 'OuteTTS-0.2-500M-Q8_0.gguf', embedding: true })
await ctx.initVocoder({ path: 'codec-q8_0.gguf' })
const { prompt, grammar, embedding, flow } = await ctx.getFormattedAudioCompletion({
  prompt: 'Hello world',
})
const result = await ctx.completion({ prompt, grammar, embedding })
const pcm = flow === 'continuous_embd'
  ? await ctx.decodeAudioEmbeddings(result.embeddings, result.embedding_dim)
  : await ctx.decodeAudioTokens(result.audio_tokens)

Voice cloning, saved voices and per-model notes: llama.rn TTS docs.

Voices

Builtin: the reference speakers of the upstream OuteTTS package (English, Japanese, Korean, Chinese); tts-cli voices lists them. No voice cloning.

Input text

Letters and spaces only: write numbers out as words; Chinese, Japanese and Korean go in romanized form (pinyin, romaji, revised romanization).

License

cc-by-nc-4.0, the same as the original model.