Audio to Text

Batch and speaker-labeled transcription using AssemblyAI.

Setup

1. Install dependencies:

pip install -r requirements.txt

2. Set your API key in a .env file (recommended):

ASSEMBLYAI_API_KEY=your_key_here

If the key is missing when you launch the GUI, a box at the top of the main window asks for it (with a text area to paste the key). You can optionally save it to .env from there.

3. Place audio files in the audio/ directory (created automatically if missing when using the GUI).

Desktop UI (Tkinter)

python3 gui.py

The app has two tabs:

• Batch — Scans audio/ for supported files, shows Pending/Done status, and transcribes files that do not yet have a matching transcripts/<name>.json.
• Speaker-labeled — Transcribes one file (from audio/ or via Browse) with speaker diarization and saves conversation.json.

Language

At the top of the window, choose how AssemblyAI should handle language:

• Auto (detect language) (default) — Uses language_detection=True (same as the CLI scripts).
• A specific language — Sets language_code manually. The dropdown lists all 103 languages from AssemblyAI supported languages (e.g. en, es, et, en_us).

With a manual language code, some features (such as speaker labels) may not be available for every language; the API returns an error in that case.

CLI

Batch transcription

python3 audio2txt.py

• Looks in audio/ for audio files.
• Checks transcripts/ for matching <same-name>.json files.
• Transcribes only files that do not yet have a matching transcript.
• Saves transcripts to transcripts/ with the same base filename and .json extension.
• Uses automatic language detection by default.

Speaker-labeled (single file)

python3 audio2txtSpeakerLabeled.py ./audio/your_file.m4a

Transcribes one file with speaker labels and writes conversation.json. Audio files live in ./audio/ by default; you can pass a filename (e.g. recording.m4a), a path under ./audio/, a full path, or a URL.