Independent developer Jodi Bruijn released the MacWhisper 14.0 update yesterday, focusing on improving transcription performance and adding an editor view. MacWhisper was launched in 2023 and is an AI-driven transcription application for Mac devices that supports running AI models such as OpenAI's Whisper and NVIDIA's Parakeet locally. It can directly handle transcription tasks for audio, video, meetings, and system audio, suitable for scenarios such as podcast production.

image.png

Interface upgrade, smoother editing experience

In terms of the interface, MacWhisper 14.0 adds an editor view, allowing users to directly edit transcripts on the main transcription interface, making it easier to add paragraphs, quickly reassign speakers, and save useful parts of the transcript. At the same time, the AI service interface has been redesigned for clearer and more user-friendly use, and updated AI models for popular AI service providers have been added. After processing multiple open text records, the playback effect is better, and the transcription window has also improved the Cmd+A operation and instant editor interaction experience.

Multiple fixes are released simultaneously with 14.1

In terms of fixes, improvements include selecting regions in Deepgram including the EU and US, controlling filler words, and expanded Nova3 language support. Cloud transcription uploads no longer load the entire file into memory, and issues such as menu bar position, audio-video synchronization, CPU idle usage, and special cases in meeting recordings have been fixed. The official also released the MacWhisper 14.1 update, which, in addition to fixing many bugs in the editor, added Gladia as a cloud transcription provider, allowing users to perform transcription tasks using their own Gladia API key.