SpaceXAI releases Grok Voice Transcribe 2.0 speech-to-text API
SpaceXAI has released Grok Voice Transcribe 2.0, a hosted speech-to-text model for batch files and real-time audio streams. The company says it is twice as accurate as version 1.0 without changing prices. The model is designed for difficult recordings, including noisy phone calls, overlapping speakers, accents and spoken credentials, and is available under model ID grok-voice-transcribe-2.0. Open weights have not been announced.
SpaceXAI reports that the model ranked first among 32 streaming systems on the public Artificial Analysis leaderboard. Its internal tests across telephony, conversations, credentials and short phrases showed improvements over version 1.0, although those results have not been independently reproduced. For short phrases in 19 languages, reported word error rate declined from 20.6% to 6.8%. The system can detect languages automatically and handle language changes within one recording.
The API includes timestamps, speaker diarization, transcription of up to eight channels, term biasing and formatting for numbers and contact details. Batch transcription costs $0.10 per audio hour, while streaming costs $0.20. Atlassian Loom has adopted the model for video transcription, according to the companies.
Written by Sakhanda Wire AI from reporting by MarkTechPost. Facts, quotes and full context are in the original article below.