Skip to main content

Superwhisper adds real-time transcription via ElevenLabs Scribe v2

By Steven Van ·

The engine streams text as you speak and supports 90+ mixed languages, free on Mac, Windows and iOS.

superwhisper, the dictation tool that turns speech into text in any app, just added a notable engine to its lineup: ElevenLabs' Scribe v2 Realtime. The pitch is simple and specific, it writes as fast as you talk, with the accuracy to keep up, which is exactly the combination dictation tools usually make you choose between.

Real-time is the feature

Most high-accuracy transcription works in a beat: you speak, you stop, the text appears. Scribe v2 Realtime streams the words in as you say them, keeping pace with natural speech without lag. For dictation, that changes the feel more than any accuracy number, you watch the sentence form while you're still mid-thought, catch mistakes immediately, and never wait on a processing spinner. As one power user quoted in superwhisper's announcement put it, that's really the main use case.

90+ languages, mixed freely

Scribe v2 Realtime supports more than 90 languages with automatic language detection, and, notably, you can mix languages within a single dictation. If you think in two languages, or drop English product terms into another language's sentences (which is how most multilingual people actually talk), the transcript follows along instead of mangling the switch.

A pragmatic turn for a local-first tool

superwhisper built its reputation on local processing: Whisper running on-device, private, offline. Adding an ElevenLabs cloud model to the roster is a pragmatic acknowledgment that different dictation jobs want different engines, local models for privacy and offline work, Scribe v2 Realtime when speed-with-accuracy matters most. Users pick per use case, and the tool stays the same everywhere it runs: Mac, Windows, and iOS, dictating into whatever app has focus.

It's also another data point on ElevenLabs' quiet expansion beyond voices. Scribe started as its transcription model; landing inside the best-known dictation utility puts it in front of people who dictate for hours a day, the harshest possible judges of latency.

Try it

Scribe v2 Realtime is available in superwhisper now, on the free tier and Pro, across Mac, Windows, and iOS. If you've tried voice-to-text before and bounced off the lag, this is the configuration to retest with: open superwhisper, select Scribe v2 Realtime as the model, and dictate a paragraph at full speed.

superwhisper
SuperWhisper
AI-powered voice-to-text that runs locally on your device — dictate anywhere with high accuracy.
View SuperWhisper →

Sources: superwhisper Blog, ElevenLabs.

Read Superwhisper adds real-time transcription via ElevenLabs Scribe v2 on Creators Toolbox