Audio quality has always been at the core of what we're building. Today we're shipping an upgrade that makes a real difference to how content sounds: Gemini Native TTS is now powering all audio on Sonacasts.
What Changed
Until now, all audio was generated using Google Cloud TTS with Chirp 3 HDvoices — solid quality, but with a ceiling. Chirp 3 HD is a traditional neural TTS model: it converts text to speech reliably, but it doesn't have a deep sense of context or natural cadence.
Gemini Native TTS is different. It's built on top of the same foundation as Gemini — a large language model that actually understands what it's reading. That means better pacing at sentence boundaries, more natural emphasis, and delivery that fits the content rather than just the words.
Hear the Difference
The same 10 voices are available. Click play to compare Chirp 3 HD and Gemini Native side by side — same voice, same text, different model:
Nothing changes in how you pick a voice. Charon, Despina, Aoede — all 10 are still there. The names are familiar, the personalities are consistent, but the delivery has a noticeably higher ceiling.
Finnish and other languages are included in this upgrade. Gemini Native handles multilingual content well, so the quality improvement applies across all supported languages.
What's Next
We're continuing to experiment with what Gemini's audio capabilities can do — including more expressive delivery for different content types: storytelling, deep dives, kids content. More on that soon.
What do you think about the audio quality?
Your feedback helps us improve the listening experience
Human in the loop
Kai Lemmetty
Co-founder of Sonacasts. Marketer turned product builder who started coding AI apps in 2024.
