TranscriptX is a local-first workbench for people who want to think with transcripts.
Import conversations you already have. See themes, speakers, and evidence. Optional local AI stays on your computer. TranscriptX does not transcribe audio — bring files from WhisperX, Scriberr, noScribe, Otter, or a similar tool.
Not sure if this is the right tool? How TranscriptX compares.
Open Charts from the same View menu for visual module outputs.
- Understand themes across a conversation
- Compare speakers — who said what, and how they interact
- Investigate a question and jump back to the original lines
- Analyse several conversations together over time
- Correct the transcript while you read
- Export findings as HTML or a ZIP you keep
Walkthroughs: Using TranscriptX. Product definition: docs/PRODUCT.md.
Source files and analysis results stay on your computer. Optional AI uses Ollama locally and stays off until you turn it on.
Limits: known limitations. Third-party models: NOTICE.
Use the sample planning_review.json if you do not have a transcript yet.
- Open Import Transcript, upload the JSON, and confirm.
- Open Speaker Identification and give each diarized label a display name (the sample uses
SPEAKER_00, … until you rename them). Most Balanced modules skip until speakers are named. - Open Run Analysis, keep Balanced, and run it.
- Open Overview and note a couple of useful outputs.
Native install from git: use ./transcriptx.sh or pip install -e ".[full,web]" so chart modules can finish (see installation).
Full walkthrough: First analysis.
Five everyday jobs: first analysis, name speakers, investigate evidence, local AI (optional), export. USB ingest can auto-name after admit: auto-identify. More: all workflows.
Docker (recommended). Copy .env.example to .env and set HOST_RECORDINGS_DIR to an absolute path outside this repository.
git clone https://github.com/glen-w/TranscriptX.git
cd TranscriptX
cp .env.example .env # set HOST_RECORDINGS_DIR
docker compose up transcriptx-webOpen http://localhost:8501. The first run builds the image.
Native (from git — not PyPI). Python 3.10–3.12. From the repo: ./transcriptx.sh creates a .transcriptx virtualenv and starts the web UI. Details: installation. Docker notes: docker. How to turn audio into a file: transcription.