Instructions to use TheStageAI/pocket-tts-100m-ru with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Pocket-TTS
How to use TheStageAI/pocket-tts-100m-ru with Pocket-TTS:
from pocket_tts import TTSModel import scipy.io.wavfile tts_model = TTSModel.load_model("TheStageAI/pocket-tts-100m-ru") voice_state = tts_model.get_state_for_audio_prompt( "hf://kyutai/tts-voices/alba-mackenna/casual.wav" ) audio = tts_model.generate_audio(voice_state, "Hello world, this is a test.") # Audio is a 1D torch tensor containing PCM data. scipy.io.wavfile.write("output.wav", tts_model.sample_rate, audio.numpy()) - Notebooks
- Google Colab
- Kaggle
pocket-tts-100m-ru — TheStage Apple SDK engines
Core ML / ANE engines of the Russian 100M-parameter Pocket TTS fine-tune
ArtShtorm/Shtorm-PocketTTS-RU by ArtShtorm (CC-BY-4.0; the 'fast' 6-layer model; INPUT TEXT MUST CARRY STRESS MARKS (combining acute accent on the stressed vowel, e.g. 'Здра́вствуйте') as in training; unstressed text is spoken with a foreign accent),
built on kyutai/pocket-tts, for the TheStage Apple SDK
(model_type: pocket-tts, bundle language ru).
subscription/macos_ios/engines.zip (143 MB) — flow LM fp16 with 6-bit g8 palettized transformer layers
(6 layers, d=1024, LSD flow head, ANE), Mimi decoder fp32 (CPU), voices: azelma (default), alba, charles, cosette, javert, marius (Kyutai's English reference clips encoded by this model; no native prompt is shipped by the author). QLPA-encrypted engines.
- Downloads last month
- -