FFTrans – Made in Japan

FFTrans Parakeet

― AI Speaker Diarization & High-Speed Transcription App for Apple Silicon ―

Offline Transcription for Maximum Privacy

---

FFTrans Parakeet is a Mac-only app that transforms important audio and video—such as meetings, interviews, lectures, and seminars—into highly accurate text in no time.
With the latest AI speech recognition engine “parakeet_mlx” and built-in speaker diarization, organizing audio files, creating meeting minutes, and editing subtitles has never been simpler or more reliable.
All processing is done locally on your Mac—no internet connection required—ensuring total confidentiality for sensitive meetings and interviews.
Automatic speaker separation (diarization) comes standard, making it easy to distinguish and record multi-person discussions or dialogues.
A simple and intuitive interface makes FFTrans ideal for both daily use and demanding business environments.

---

Make all your voice work smarter

• Automatically label multiple voices (speaker separation) for easy meeting minutes organization.
• Automatically remove unnecessary and repeated words for dramatically reduced editing effort.
• Support for time-stamped SRT makes video subtitle creation quick and easy.
• Intuitive and simple UI Easy to use, even for beginners

A diverse and easy-to-use set of features

• Improve transcription reliability with proprietary AI technology that automatically removes hallucinations (false positives) and unnecessary repetitions.
• Equipped with a function to split sentences at periods (.) in speech, supporting natural text layout.
• Reliably processes long audio recordings, even those over two hours long. Securely record entire recordings and meetings with segmented analysis.

Get Started Now

System Requirements

• Compatible Mac: Apple Silicon (M1/M2, etc.) models only (will not work with Intel Macs)
• macOS Version: 15 (Sequoia) or later recommended
• Memory: 8GB or more (16GB or more recommended) (Please close other apps as much as possible to free up memory)
• Free Disk Space: Please ensure you have sufficient free space for saving and analyzing audio files

Version History

[ Version 1.4 ]
• Add English Priority Mode
• Change Icon to macOS Tahoe’s Squall
• Added processing that skips too short audio processing with speaker separation

[ Version 1.3 ]
• The maximum number of speaker registrations to 100 people (previously 30 people)
• Keep the SRT display position when custom dictionaries and speaker names are reflected
• Change transcription processing from thread to process processing to ensure memory after transcription completion
• Added processing to release MLX cache sequentially using the memory API metal memory

FFTrans offers industry-leading accuracy, robust privacy, and powerful support for English and Europian languages and long audio—making any audio workflow faster and more secure. Try it today!

---