SpeechFast
A local voice input application for Mac that performs offline speech-to-text based on the Whisper model, supporting global hotkey input, text cleanup, and custom vocabularies.
Timeline (last 90 days)
The interaction method of SpeechFast is: hold down the fn key and speak, and after releasing, the text automatically appears at the current cursor position, supporting use in any application where text can be entered.
SpeechFast's speech model only needs to be downloaded once, and thereafter it can work offline, making it suitable for network-free environments such as on airplanes or within closed enterprise intranets.
SpeechFast uses Apple Intelligence to clean up filler words and slips of the tongue (such as "um" and "you know"), outputting cleaner text.
SpeechFast provides a custom vocabulary feature, where users can add proper nouns, personal names and industry terms to improve recognition accuracy.
SpeechFast provides a locally saved dictation history, and the history records exist only on the local machine.
SpeechFast is not suitable for users who pursue the highest recognition accuracy or need advanced features such as multi-speaker separation and long-audio transcription.
SpeechFast's disadvantage is that its recognition capability is limited by the local model and lacks the continuous large-model iteration of cloud services.
All Facts (7)
The interaction method of SpeechFast is: hold down the fn key and speak, and after releasing, the text automatically appears at the current cursor position, supporting use in any application where text can be entered.
50%UnverifiedSpeechFast's speech model only needs to be downloaded once, and thereafter it can work offline, making it suitable for network-free environments such as on airplanes or within closed enterprise intranets.
50%UnverifiedSpeechFast uses Apple Intelligence to clean up filler words and slips of the tongue (such as "um" and "you know"), outputting cleaner text.
50%UnverifiedSpeechFast provides a custom vocabulary feature, where users can add proper nouns, personal names and industry terms to improve recognition accuracy.
50%UnverifiedSpeechFast provides a locally saved dictation history, and the history records exist only on the local machine.
50%UnverifiedSpeechFast is not suitable for users who pursue the highest recognition accuracy or need advanced features such as multi-speaker separation and long-audio transcription.
50%UnverifiedSpeechFast's disadvantage is that its recognition capability is limited by the local model and lacks the continuous large-model iteration of cloud services.
50%