dlu-voice-transcriber

DLU Voice Transcriber

A Python tool for recording audio and transcribing it to text in multiple languages (English and German).

Features

• High-quality audio recording (48kHz sample rate, 16-bit depth)
• Automatic input device detection and selection
• Support for multiple audio formats (WAV, MP3)
• Configurable MP3 quality settings
• Transcription to English and German
• Progress monitoring during recording
• Save transcriptions to text files
• Detailed device information and listing
• Intelligent device selection with native sample rate support

Installation

pip install dlu_voice_transcriber

System Requirements

• Python 3.8 or higher
• Working microphone
• Linux: sudo apt-get install python3-pyaudio portaudio19-dev
• macOS: brew install portaudio
• Windows: No additional requirements

Usage

Basic Commands

List available audio devices:

dlu_transcribe --list-devices

Basic recording (5 seconds, default settings):

dlu_transcribe

Advanced Usage

Record with specific settings:

dlu_transcribe --device 2 --duration 10 --output my_recording --format mp3 --mp3-quality 320 --language both --save-text

Command Line Options

• --device: Specify input device index (shown by --list-devices)
• --duration: Recording duration in seconds (default: 5)
• --output: Output filename without extension (default: recording)
• --format: Output format: 'wav' or 'mp3' (default: mp3)
• --mp3-quality: MP3 quality in kbps (default: 320)
• --language: Transcription language: 'de', 'en', or 'both' (default: both)
• --save-text: Save transcriptions to text files
• --list-devices: Show available audio input devices

Examples

List all available devices:

dlu_transcribe --list-devices

Record high-quality MP3 for 15 seconds:

dlu_transcribe --duration 15 --format mp3 --mp3-quality 320 --output high_quality_recording

Record using specific device with German-only transcription:

dlu_transcribe --device 2 --language de --save-text --output german_speech

Output Files

The tool generates the following files in the recordings directory:

• Audio file: <output>.<format> (e.g., recording.mp3)
• Transcription files (if --save-text is used):
  • English: <output>_en.txt
  • German: <output>_de.txt

License

This project is licensed under the MIT License - see the LICENSE file for details.