Piper or Kokoro: which text-to-speech model to pick
Both models are tiny and run on a regular CPU, no GPU needed. Piper targets hardware down to a Raspberry Pi, ships ready-made voices in 35+ languages including several Russian ones, and covers notifications, voice menus and offline kiosks. Kokoro is 82M, fully Apache 2.0 with no licensing homework, but it has no Russian. Piper's licensing is messier: the old engine is MIT, the new piper1-gpl is GPL-3.0, and every voice carries its own terms, some of them non-commercial, so each voice has to be checked. If you add voice cloning on top of either one, remember that a voice may only be cloned with the consent of the person it belongs to.
Comparison based on catalog data
| Parameter | Piper | Kokoro |
|---|---|---|
| Category | Text to speech | Text to speech |
| Developer | Rhasspy / Open Home Foundation, USA | hexgrad (independent developer), not disclosed |
| Releases | Jun 2023 – Mar 2025 | Dec 2024 – Feb 2025 |
| Sizes | about 5M – 30M | 82M |
| Hardware | Laptop | Laptop |
| Commercial use | Commercial use with conditions | Commercial use allowed |
| License | The old engine is MIT, the new piper1-gpl is GPL-3.0; each voice has its own license (some free, some non-commercial) | Apache 2.0 |
| Russian | Supported | Not supported |
| Ollama | No | No |
| Without GPU | Yes | Yes |
| Tasks |
|
|
Choose Piper if
- You need Russian: Piper has ready-made Russian voices, Kokoro has none
- Speech runs on weak hardware such as a Raspberry Pi or a kiosk
- Smart home, embedded devices and offline voice menus
Choose Kokoro if
- You want simple licensing: Kokoro is entirely Apache 2.0
- English voiceover for a web or mobile app without a GPU
- You do not want to check terms for every individual voice
Other comparisons
- Silero or Piper: which to choose for business
- XTTS or F5-TTS: which voice cloning model to pick
- F5-TTS or Fish Speech: which voiceover model to pick
- GigaAM or Vosk: which speech recognition to pick
- Qwen or GigaChat: which to choose for business
- Qwen or Llama: which to choose for business
- GigaChat or YandexGPT: which to choose for business
- DeepSeek or Qwen: which to choose for business
- Gemma or Llama: which to choose for business
- Mistral or Llama: which to choose for business
- Whisper or GigaAM: which to choose for business
- Whisper or Parakeet: which to choose for business
Need a model for your task?
An open model can run on your own server: data stays in-house, there is no per-request fee, and the model can be fine-tuned on your documents.
- SelectThe model and size for your task and hardware budget
- DeployOn your server or in a closed network, with an API
- Fine-tuneOn your data, or connect a knowledge base
- IntegrateInto your CRM, ERP, bot, website or team chat