User guide
Create your first subtitles.
Install the app, prepare a Whisper model, and import an audio or video file.
Install the app
Check the download center for supported platforms and version status. Download ↗
macOS Apple Silicon
After obtaining a release package, open the DMG and drag the app to Applications, or extract the ZIP and move the app to Applications.
Windows x64
Run the EXE installer, or fully extract the ZIP and launch the Toucan Subtitle executable. Install or repair the Visual C++ x64 runtime if Windows reports it missing.
Microsoft runtime download ↗Verify files
Replace the example file name in the command with yours and compare the result with the SHA-256 list from the same GitHub Release.
macOS
shasum -a 256 "installer.dmg"Windows PowerShell
Get-FileHash "installer.exe" -Algorithm SHA256Prepare models
- Open Settings → Speech transcription and select the engine and model size.
- Download the transcription model and wait for the weights to finish downloading.
- Open Health Check to confirm the runtime, engine and weights are ready.
Set up a model for polishing or translation
No large language model is bundled with the app. To polish or translate subtitles, install Ollama and download a model, or supply your own API key for an OpenAI-compatible cloud service.
For local Ollama, enter the model name. For an OpenAI-compatible service, enter the base URL, API key and model name. Transcription alone requires Whisper.
Transcription uses separately downloaded Whisper weights and requires neither Ollama nor an API key.
Generate subtitles
- Return to the workspace and add files or folders.
- Select transcription only, source polishing or subtitle translation. Pick a target language for translation.
- Start the batch and check its progress. Confirm the actual source language if prompted.
- Review the subtitle preview, then find the SRT files next to the media.
Data and privacy
Local transcription
The app reads the media, extracts the audio and runs Whisper on your computer. The initial download of model weights needs an internet connection.
Where polishing and translation run
The app sends subtitle text to the endpoint you configure. Local Ollama processes it on your computer. Remote services apply their own processing and retention policies.
Website demo
The homepage uses a fixed sample and has no media upload.
FAQ
Do I need Ollama to generate source-language subtitles?
Select transcription only and download the required Whisper weights.
What if Windows reports a missing runtime?
Check the Visual C++ x64 runtime. The app folder includes support/diagnose-engine.cmd to check engine dependencies.
How do I resume an interrupted job?
Reopen the workspace and check the restored queue. The app reuses compatible checkpoints. If the model, source language or file has changed, follow the prompt to resume or restart.
What should I check when a job finishes with a warning?
Review quality flags, uncertain segments, entries still in the source language and important dialogue.
How do I report a problem?
Include your system architecture, app version, job stage and reproduction steps. Remove media names, paths, subtitle text and keys before sharing logs. Feedback & Help ↗ · GitHub Issues ↗
Batch convert videos to SRT offline
Choose Transcription only in Toucan Subtitle and download the Whisper weights. You can then queue a folder of videos locally and export timecoded SRT files. This workflow needs no Ollama model or API key. User guide ↗
Translate subtitles with Ollama or your own API key
Toucan Subtitle transcribes the media locally with Whisper, then uses a model you configure to translate the subtitle text. The app retains the transcription timeline and matches the translated text to its cues. The installer does not bundle a large language model. User guide ↗