Fast, accurate, private.
Sona keeps speech-to-text on your own machine, then lets you review timestamps, clean up wording, translate, and export without leaving the same editor.
Workflow
Product Preview
Sona keeps the capture, timestamps, polish, translation, and export path inside one working surface.
Stage
The transcript arrives first, with timing intact and almost no chrome between the audio and the page.
Let us keep the local recording path for the next release and tighten the onboarding copy once the build settles.
And keep the timestamps attached so reviewers can jump back to the exact moment in the audio.
Once the first pass is done, run AI polish so the draft reads cleaner before the team sees it.
For external collaborators, generate a Chinese version too, but keep the original English visible beside it.
That way we can compare wording quickly without splitting the transc...|
Use Cases
Sona handles local transcription first, then keeps polish, translation, and export in the same working thread instead of sending you across separate tools.
Context
Capture project syncs, interviews, or standups while the room is still talking.
Workflow
Use Live Record to keep timestamps attached, then run polish for a cleaner internal recap.
Result
You leave with a shareable draft that still points back to the original audio moments.
Context
Record lectures, seminars, or office-hour explanations without losing the structure of the talk.
Workflow
Transcribe locally, jump back through timestamps, and edit key segments into study-ready notes.
Result
The transcript stays tied to playback, so review and revision happen in one place.
Context
Bring in local video or film files when you need subtitles instead of a blank timeline.
Workflow
Queue them through Batch Import, review the text, then export SRT or VTT once the segments look right.
Result
You get subtitle files that are ready for handoff, iteration, or platform upload.
Context
Prepare subtitles for bilingual review or international delivery without separating the source text.
Workflow
Keep the original transcript visible, run Translate, and export translation-only or bilingual output when needed.
Result
The translated version stays aligned with the source so subtitle comparison is quick and readable.
Local transcription works on-device; polish and translation require a configured LLM provider.
Local-first by default. Transcription runs on your hardware with Sherpa-onnx; optional downloads, updates, sync, and provider-backed AI actions connect only when you choose them.
Built-in capabilities with local Large Language Models to automatically polish, summarize, and translate your transcripts.
Achieves commercial-grade speech recognition tailored for long audio files with intelligent phrasing.
Rich, interactive editor built with React to easily tweak generated text, align timestamps, and refine your audio.
Start Here
Install the latest release, then record or import audio, review timestamps, polish, translate, and export without breaking the thread.
Windows, macOS, and Linux builds are pulled from the latest GitHub release.