Whisper Transcription

This app brings the power of the Whisper AI model to the Mac desktop. The transcription accuracy is great, handling technical jargon, multiple speakers, and challenging audio quality with surprising precision. The experience is perfectly tailored for macOS, feeling like a native utility. Drag-and-drop functionality for importing media is seamless, and the transcription speed (even when using larger models) is impressively fast on Apple Silicon. The ability to process files locally on my Mac without sending sensitive audio to the cloud is a huge advantage for privacy and security. Although my workflow has a few niche requirements that the current set doesn't completely address, the version on the App Store is more than adequate for most people’s needs. The developer is very active in the community and is constantly making improvements to the app. If you deal with audio or video files and need a solid transcription on your desktop, look no further.

Good local option. It’s kind of like a feature-limited version of Descript. I purchased the Pro license for larger models. Even with M4 Max, it’s not as fast as cloud services, but overall works well and is more functional than the open source project. Recommended. I liked the option to combine segments by sentence, but found speaker identification needs a lot of improvement (it frequently associates one word in a sentence to another/wrong speaker or the last/first word of a sentence from the previous/next speaker). I was also a bit disappointed there is no CLI/wrapper to use it from the terminal. I was able to figure out how to filter out filler words from the transcript, but it would be nice if there was a one-click way to enable doing so. Please keep improving.

Mouse and keyboard are out - voice is in. The current version of Whisper Transcription is half voice transcriber, half LLM. That opens up many unique workflow capabilities such as: 1) record yourself babbling into a voice recorder, upload the audio file into W.T. and use the ChatGPT API to extract the key points and action items from your monologue (this is the BEST way to get organized at the start of your busy day); 2) record your Zoom/Teams/etc. meeting with permission and upload the video or audio file(s) into W.T., then use the ChatGPT API to create recap meeting notes and action items (gone are the times of missing key points because you dissociated for 15 mins) [note: being able to do this all on-device and securely would be the ultimate selling point]; 3) paste a video link and get the full transcription without having to sit through the video, then use the ChatGPT API to extract key points or a summary and decide if the video is worth watching through fully. Great work - huge fan - absolutely recommend. This is how we will all interact with computers from here on out.