Grok Voice Is Live on fal for Low-Latency Voice Agents
Key Info
Grok Voice is now live on fal, giving developers access to a speech model designed for intelligent, low-latency voice agents. It responds in 0.70 seconds, completes tool calls before the spoken sentence ends, supports word-level timestamped transcription, and offers TTS in 30+ voices across 25+ languages.
Highlights
- Build low-latency voice agents on fal for real customer-service use cases.
- Responses come in 0.70 seconds, with tool calls finishing before the sentence ends.
- Includes word-level timestamped transcription and TTS in 30+ voices across 25+ languages.
- Voice can be cloned from just two minutes of audio; every voice in the announcement video is generated with Grok Voice.