xAI opens Grok STT and TTS audio APIs, with overall word error rate of STT reduced to 6.9%

ME News message, April 18 (UTC+8). According to Beating monitoring, xAI has launched two standalone audio APIs: Grok Speech to Text and Grok Text to Speech. Both come from the same audio stack that supports Grok Voice, Tesla’s in-car system, and Starlink customer service. This time, they are opened as independent endpoints, allowing developers to directly integrate into applications such as voice agents, real-time transcription, accessibility tools, and podcasts.

STT provides two modes. The REST API is for batch transcription of large audio files, returning results at millisecond-level latency; the WebSocket API is designed for real-time audio streams. Included capabilities include word-level timestamps, speaker separation (diarization), multi-channel recognition, and Inverse Text Normalization, which automatically transforms spoken numbers, dates, and currencies into standardized structured text. Language coverage exceeds 25, with seamless switching during conversations.

xAI also released a set of word error rate (WER, lower is better) comparisons. In overall scenarios: Grok 6.9%, ElevenLabs 9.0%, Deepgram 11.0%, AssemblyAI 12.9%. The gap is even larger in “telephone call entity recognition”: Grok 5.0%, while the corresponding figures for the other three are 12.0%, 13.5%, and 21.3% respectively. In common business scenarios such as meetings, video podcasts, and phone calls, Grok also maintains a slight lead. These figures are published based on xAI’s own testing, and there has been no third-party re-verification yet.

In pricing, batch STT is $0.10 per hour, and streaming STT is $0.20 per hour; TTS is $4.20 per 1 million characters. TTS supports inline Speech Tags to control emotion and prosody, such as \[laugh\], \[sigh\], \[whisper\], and others.

(Source: BlockBeats)