Generates a WhatsApp-style HTML page from an exported WhatsApp chat, with support for images, videos, audio, PDFs, and optional audio transcription.
pip install -r requirements.txt
For audio transcription, set your OpenAI API key:
# Linux/macOS
export OPENAI_API_KEY="sk-..."
# Windows
setx OPENAI_API_KEY "sk-..."
python whatsapp_viewer.py --dir "path/to/chat/folder"
This expects a chat.txt file and media files inside the folder, and generates output.html there.
python whatsapp_viewer.py "my_chat.txt" --dir "path/to/folder" --me "YourName"
The --me flag aligns your messages to the right.
Transcribe all audio files using OpenAI’s speech-to-text API:
python whatsapp_viewer.py "chat.txt" --dir "path/to/folder" --transcribe
Transcribe only the first N audios (useful for testing):
python whatsapp_viewer.py "chat.txt" --dir "path/to/folder" --transcribe --transcribe-only-x-audios 5
Transcriptions are cached as .original.txt files next to each audio file. Re-running the command skips already-transcribed audios.
Correct transcriptions using an LLM with conversation context:
python whatsapp_viewer.py "chat.txt" --dir "path/to/folder" --correct
Interactive mode lets you review each correction (accept, reject, or edit):
python whatsapp_viewer.py "chat.txt" --dir "path/to/folder" --correct-interactive --transcribe-only-x-audios 10
Corrected transcriptions are saved as .txt files. The original .original.txt files are preserved. The HTML output uses corrected versions when available.
| Argument | Description |
|---|---|
chat_txt |
Chat text file (default: chat.txt) |
media_dir |
Media directory (default: .) |
output_html |
Output HTML file (default: output.html) |
--dir DIR |
Base directory for all files |
--me NAME |
Your name in the chat (right-aligns your messages) |
--transcribe |
Transcribe audio files using OpenAI API |
--transcribe-only-x-audios N |
Limit to first N audios |
--stt-model MODEL |
Speech-to-text model (default: gpt-4o-mini-transcribe) |
--correct |
Correct transcriptions using LLM with conversation context |
--correct-interactive |
Interactively review each correction |
--llm-model MODEL |
LLM model for correction (default: gpt-4o-mini) |
A sample/ directory is included with a short conversation between two fictional users, AI-generated images, audio files, and pre-generated transcriptions.
To generate the HTML yourself:
python whatsapp_viewer.py --dir sample --me Bob
Or browse the pre-generated output directly: