Transcribe audio

Before you start

Authenticate using the installation guide. Prepare a recording you are authorized to process at a URL the server can access. The CLI does not upload local files for transcription. A localhost URL is not reachable from the remote service. Do not expose confidential recordings merely to obtain a public URL.

Inspect and submit

kitta audio transcribe --schema
kitta audio transcribe --audio-url https://example.com/recording.wav
kitta audio transcribe --audio-url https://example.com/recording.wav --language en

Replace the example URL with your recording. These are two alternative paid requests: run only one. Omitting --language uses automatic detection. Inspect the schema for accepted language values and current request constraints.

Save the JSON result

kitta audio transcribe --audio-url https://example.com/recording.wav --json > transcript.json

This submits another transcription; it does not retrieve the previous result. stdout contains JSON rather than an automatically generated subtitle file. Check the exit code before consuming transcript.json; redirection may create an empty file on failure.

Failure and recovery

An inaccessible URL or unsupported audio may fail. A timeout is an unknown outcome, not proof that the service did no work. Transcription is not automatically retried. Inspect account usage before resubmitting. TTS jobs commands cannot recover transcription jobs.

Next: automation and billing.