elevenlabs-stt

To transcribe a podcast with speaker diarization and timestamps, this Skill processes common audio formats and outputs a JSON transcript containing word-level timestamps and speaker labels. You need an ElevenLabs API key and command-line tools like curl and jq for requests and parsing.

The best way to convert multi-speaker meeting recordings into time-aligned JSON transcripts is using an automated speech-to-text process with speaker diarization. This automatically identifies individual speakers and maps spoken words to precise timestamps for downstream summarization or note-taking.

Yes, you can use curl and jq for speech-to-text transcription of interviews across multiple languages. This Skill requires curl to send audio file requests to the ElevenLabs API and jq to parse the resulting JSON output, supporting over 90 languages with automatic or explicit language selection.

Speech-to-text processing does support audio event tagging for non-speech sounds like laughter and music. During transcription, the audio processing mechanism detects these specific events and includes them alongside word-level timestamps and speaker labels in the final JSON output.

Yes, you need an ElevenLabs API key to generate word-level timestamps from voice notes. The transcription process sends audio data to the ElevenLabs API via curl, returning a JSON transcript with precise time-alignment and optional speaker diarization that requires jq for local parsing.

A limitation of using command-line tools like curl and jq for audio transcription and diarization is the lack of a graphical interface, requiring manual terminal execution. Transcription outputs are raw JSON, meaning users must handle formatting text and speaker labels themselves for readable transcripts.