Speechify’s Windows app uses local models for transcription and dictation

In Brief

Voice AI institution Speechify conscionable launched a autochthonal Windows app that employs locally stored models to alteration dictation crossed apps, and speechmaking aloud articles, documents, oregon PDFs utilizing its room of voices.

The institution is taking connected the likes of Wispr Flow, Willow, and Superwhisper who besides supply dictation and transcription apps crossed platforms.

Speechify said Windows app does dependable processing wholly on-device connected Copilot+ PCs (that person NPUs from AMD, Intel, and Qualcomm) and different Windows 11 PCs that person GPUs from Intel and AMD.

The app has 3 models moving on-device: neural text-to-speech, real-time dependable enactment detection, and Whisper-powered transcription. Users tin configure the app to power to cloud-based models oregon adjacent alteration them during usage.

The company, which has implicit 50 cardinal users, said that VITS Neural tin make audio crossed 7 antithetic velocity presets, allowing users to person the app work aloud documents oregon web pages. The institution uses the Silero open-source model for dependable enactment detection.

“Over a cardinal radical connected this satellite usage Windows. With this Windows launch, we’re making definite that reading, and present writing, is ne'er a barrier, nary substance what instrumentality you usage oregon however you similar to work. We’re particularly excited astir the accidental successful the endeavor fixed however galore professionals person asked for Speechify connected their PCs,” said Cliff Weitzman, laminitis and CEO of Speechify, successful a statement.

Last month, the institution launched Granola-like gathering transcription, but that diagnostic was constricted to browser-based meetings. Now that the institution has apps crossed platforms, it volition apt bring implicit this diagnostic to autochthonal apps to transcribe meetings connected immoderate app oregon browser.

Until a fewer years ago, Speechify mostly concentrated connected text-to-speech usage cases specified arsenic speechmaking retired articles and emails, and generating podcasts retired documents. Lately, the institution has been trying to go a full-stack dependable app for users by launching dictation, gathering transcription, and a dependable assistant.

Subscribe for the industry’s biggest tech news