Google released two new text-to-speech models today, named gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts. According to reports from Simon Willison, the release includes a library containing over two thousand voices. Additionally, the system allows users to generate a custom voice by providing a thirty-second audio sample, provided they own the rights to the voice material. Following the release, a developer built a bring-your-own-key playground interface utilizing GPT-6 Astra, taking advantage of the open CORS policy present in the underlying Gemini API.
Expanding Voice Capabilities in AI
Text-to-speech technology has increasingly focused on hyper-realism and customization, moving from robotic synthesized speech to models capable of replicating human intonation and timbre. Google’s introduction of these specific flash-tier models points toward a strategy of providing fast, low-latency audio generation while integrating deeply with their existing Gemini infrastructure. The inclusion of custom voice cloning via short audio samples reflects a broader industry push towards personalized synthetic media generation, though it also raises technical and operational questions regarding access controls.
Unanswered Questions and Future Watch
Despite the initial details shared following the launch, several technical specifics remain unaddressed. The announcement did not specify the exact pricing structure for API usage of gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts, nor did it outline the safety guardrails implemented to prevent the unauthorized cloning of voices without consent. Furthermore, it remains unclear whether the open CORS policy observed by developers was intended for production environments or if it was an oversight during the rollout. Developers and users will need to monitor how Google addresses authentication and security for these endpoints moving forward.
As developers continue to experiment with custom interfaces for the new text-to-speech models, official documentation from Google is expected to clarify API limits and broader availability.
Lead image: Alpha En / Pexels
