by ElevenLabs
ElevenLabs Scribe is a speech-to-text service that transcribes recorded audio and video, and live speech, in 90+ languages. It is available through an API and SDKs, and in ElevenLabs Studio for captions and editable transcripts.
APIUsage-based
Step 1: Sign up for ElevenLabs and open Studio, or get an API key for the Speech to Text API.
Step 2: Upload an audio or video file, or stream live audio for realtime transcription.
Step 3: Add key terms to help Scribe spell names and jargon correctly.
Step 4: Turn on options such as speaker labels, audio tagging or PII redaction.
Step 5: Review and edit the transcript, then export captions or subtitles.
The pages state 99 languages for transcription, with realtime transcription in 90+ languages. Accuracy varies by language.
Yes. The realtime variant transcribes live audio with latency of about 150 ms over WebSockets.
You can upload audio or video in formats such as MP4, MOV, MP3 and WAV. The pricing table lists a 3 GB maximum file size.
HIPAA customers must contact ElevenLabs sales to sign a Business Associate Agreement before HIPAA-related deployments.
CartesiaCartesia is an AI lab that offers real-time text-to-speech, speech-to-text and a platform for building voice agents, all through an API.
OpenAI Realtime APIOpenAI's Realtime API lets developers build speech-to-speech voice agents that work directly with audio, keep conversation state and call tools. It is for developers building voice assistants and other low-latency conversational apps.
Inworld TTSInworld TTS is a realtime text-to-speech API with voice cloning and voice design, aimed at consumer apps such as companions, tutors and games. It returns first audio in under 100 ms.