Command: /transcribe
Relevant source files
The following files were used as context for generating this wiki page:
The /transcribe command is a feature of the voice-elevenlabs Claude Code plugin that converts speech from an audio file into text. It utilizes the ElevenLabs Scribe v2 model to provide fast and accurate transcriptions. This command is one of several voice-related tools in the plugin, which also includes text-to-speech capabilities like /speak and /dialogue.
Sources: commands/transcribe.md:1-5, README.md:32, README.md:43
Features
The /transcribe command leverages ElevenLabs Scribe v2 to offer several key features for high-quality transcription.
| Feature | Description |
|---|---|
| High Accuracy | Employs state-of-the-art speech recognition for precise text conversion. |
| Multiple Formats | Supports a wide range of audio formats, including MP3, WAV, M4A, FLAC, OGG, and WEBM. |
| Speaker Diarization | Can identify and label different speakers within the audio. |
| Timestamps | Provides word-level timing information for easy navigation and reference. |
| Language Detection | Automatically detects and transcribes over 70 languages. |
Sources: commands/transcribe.md:8-14
Usage
To use the command, you must provide a source for the audio file, which can be either a local file path or a publicly accessible URL.
Syntax
From a local file:
/transcribe file:/path/to/audio.mp3
From a URL:
/transcribe url:https://example.com/audio.mp3
Sources: commands/transcribe.md:18-25
Command Workflow
The command follows a simple workflow from user input to final output. It parses the provided argument, calls the ElevenLabs API, and then formats the result. An ELEVENLABS_API_KEY environment variable must be configured for the API call to succeed.
Sources: README.md:18-21, commands/voice-config.md:13
This diagram illustrates the data flow for the /transcribe command.
Sources: commands/transcribe.md:39-43
While the exact API call for /transcribe is not detailed, other commands in the plugin like /voices demonstrate the pattern of using curl to interact with the ElevenLabs API.
Sources: commands/voices.md:21-23
Input Parameters
The command accepts one of two possible arguments, specified with a prefix.
| Argument | Description | Example |
|---|---|---|
file | The local file path to the audio file to be transcribed. | file:~/Downloads/meeting.mp3 |
url | A URL pointing to a remote audio file. | url:https://podcast.com/episode.mp3 |
Sources: commands/transcribe.md:45-46
Output Format
The transcription output is comprehensive, providing more than just the raw text.
- Full Text Transcript: The complete transcribed text from the audio.
- Speaker Labels: Identifiers for each speaker if diarization is successful.
- Timestamps: Timing information for each spoken segment.
- Confidence Scores: A metric indicating the model's confidence in the accuracy of the transcription.
Sources: commands/transcribe.md:28-32
Use Cases
The /transcribe command is suitable for a variety of applications, turning spoken content into searchable and readable text.
- Meeting Notes: Automatically transcribe recordings of meetings.
- Voice Memos: Convert quick audio notes into text for easy reference.
- Podcast Editing: Generate a searchable transcript to aid in the editing process.
- Accessibility: Create captions from audio content for users with hearing impairments.
Sources: commands/transcribe.md:34-37