Desktop

Name speakers on local transcripts

OpenASR Desktop Voice ID lets you enroll a voice and attach a name to that speaker on future local file transcripts. It supports every local ASR model: Desktop keeps native speaker tracks when available and uses a shared external diarization pipeline otherwise. Voice ID is a local labeling convenience, not authentication or identity verification.

Where it runs
OpenASR Desktop on macOS and Windows; processing stays local by default
What you enroll
A voiceprint tied to a display name for that speaker on the transcript
Speaker source
Native tracks when available; shared external diarization otherwise
What it is not
Not login, access control, or biometric authentication

How Voice ID fits a meeting or interview

Multi-person audio is first separated so turns can be attributed to distinct speakers. When a voice matches an enrolled voiceprint, Desktop can show the name you assigned instead of a generic speaker label.

That makes review faster when the same people recur across recordings—clients, colleagues, or recurring interview subjects—without sending audio to a remote transcription service.

Local by default

OpenASR is local-first speech-to-text. Desktop does not require an account, does not phone home telemetry, and does not route ordinary transcription through a central cloud backend.

Models come from a signed catalog. Normal transcription does not silently auto-download models; you choose what is installed and used on the machine.

Every local file ASR model

The Voice ID switch works with every local ASR model in Desktop file transcription. MOSS currently supplies native speaker tracks; other models use FireRedVAD, segmentation-3.0, ReDimNet2-B6, and automatic clustering behind the same user flow.

Voice ID is intentionally unavailable in Desktop Dictation, Live Captions, and Remote Compute.

  • Enroll a voiceprint and assign a display name
  • Enable the same Voice ID switch with any local file ASR model
  • See named speakers when matching evidence is sufficient
  • Keep enrollment and transcription on-device

CLI and API counterparts

For automation outside Desktop, the open-core CLI exposes anonymous --diarize and the local HTTP server exposes operator-only /v1/voice-id/* identity routes plus reviewed history speaker assignments.

Desktop Voice ID is the product-facing enrollment and naming experience; the open core remains Apache-2.0 Rust CLI plus local HTTP API for scripted workflows.

Frequently asked questions

Is Voice ID the same as logging in or verifying identity?

No. Voice ID is a diarization convenience: it helps label speakers on a transcript from enrolled voiceprints. It is not authentication, access control, or proof of legal identity.

Do I need the cloud for Voice ID?

No. OpenASR Desktop runs Voice ID and speaker separation locally by default. There is no account requirement and no central cloud transcription backend for ordinary use.

Which platforms support Desktop Voice ID?

OpenASR Desktop is available for macOS and Windows. There is no native Linux desktop app; Linux users can use the open-core CLI, local HTTP API, or container images instead.

Do all models support Voice ID?

Every local ASR model supports Voice ID in Desktop file transcription. A Built-in speaker separation badge only identifies native speaker tracks; it is not a compatibility limit.

Can I do speaker work without Desktop?

Yes. The open-core CLI supports anonymous --diarize, and the local server exposes operator-only /v1/voice-id/* routes for identity workflows.

Speaker diarization app Local meeting and interview separation for every file ASR model, with optional named speakers. Diarization docs CLI --diarize, Desktop Voice ID, and the /v1/voice-id/* API. Desktop Voice ID release What shipped with Desktop Voice ID and how enrollment fits the workflow. Download Get OpenASR Desktop for macOS or Windows.