Writing on speech infrastructure
Notes from the team on voice AI, audio pipelines, and building production-ready speech products.
- Speechbase Team
Which TTS Provider Should You Use?
A practical comparison of 15 text-to-speech providers: what each one is good at, what it costs, and what to test before you ship.
- Pierson Marks
Using MiniMax Speech 2.8 for Multi-Speaker Conversations
MiniMax Speech 2.8 is now available in Speechbase and an official provider in the open-source SpeechSDK. Swapping to MiniMax is a single model string change, and with the SpeechSDK, you can easily mix MiniMax with other provider voices in the same conversation.
- Speechbase Team
Writing a WSOLA time-stretcher for the Speech SDK
Pitch-preserving speed control on speech audio, in TypeScript, with no dependencies. Why we wrote it from scratch and the bugs we hit getting it right.
- Speechbase Team
Building the Universal Speech SDK
Why we built one unified SDK for every voice model, and how it makes shipping production-ready speech dramatically simpler.