The founders' 25 MB CPU speech model remains part of the lineup. KittenML now also offers a larger speech model, hosted text-to-speech and transcription APIs, and an emotion-aware transcription product.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [Y Combinator](https://www.ycombinator.com/companies/stellon-labs)
Why it matters #
Stellon has turned a small open-source speech model into one part of a broader voice-AI offering that includes larger models and paid hosted APIs. The available adoption figures measure developer attention, while published prices show the commercial offer; neither shows whether the business has recurring customers.
Stellon Labs, whose product brand is KittenML, has expanded beyond the 25 MB speech model that introduced the company. Co-founders Rohan Joshi and Divam Gupta now offer KittenTTS 2, a 1.7-billion-parameter speech model, hosted text-to-speech and transcription APIs, and Kitten-ASR Pro, an emotion-aware transcription product.…