Show HN: Three new Kitten TTS models – smallest less than 25MB (github.com)

Article summary

Kitten TTS is an open-source, lightweight text-to-speech library that delivers high-quality voice synthesis on CPU without requiring a GPU. The library has models ranging from 15M to 80M parameters, with the smallest model being less than 25MB. It supports adjustable speech speed, text preprocessing, and 24 kHz output. The library is currently in developer preview, with commercial support available for integration assistance, custom voices, and enterprise licensing.

Main themes

• Text-to-Speech Technology
• Open-Source Library
• CPU Optimization
• Custom Voices
• Commercial Support
• Mobile Deployment

What commenters say

• The library's small size and CPU optimization make it suitable for edge deployment and mobile devices.
• Some users are interested in using the library for business applications, such as automated outgoing calls, but require more professional-sounding voices.
• The library's current lack of support for expressive tags and intonation control is a limitation for some users.
• The use of the library for automated calls raises concerns about the potential for annoying or deceptive uses of the technology.
• Some users have experienced difficulties with setting up the library due to Python version compatibility issues and environment setup.
• The developer is responsive and open to feedback, with plans to add support for custom voices, multilingual models, and improved expressive control.
• The library's open-source nature and small size make it an attractive option for developers looking for a lightweight text-to-speech solution.
• Some users suggest using tools like Conda or Nix Flakes to simplify the setup and dependency management process.