Transcribe audio and video to text on your computer

Drag a file into the Library: MP3, M4A, WAV, AAC, OGG or FLAC audio, or MP4, MOV, MKV or WebM video. Glimpse pulls the audio out and transcribes it. The transcript follows along as the audio plays, and clicking a paragraph plays from there, so checking a quote takes seconds.

On an M2 Pro MacBook, Parakeet transcribed a 9 minute recording in under 2 seconds. Whisper Large V3 Turbo took 13 seconds on the Mac and under 8 on a Windows PC with an NVIDIA graphics card. No graphics card is needed: on a 24-core processor alone, Parakeet did the same recording in about 24 seconds. See every benchmark.

Interviews, podcasts and meetings come back split by speaker. NVIDIA Nemotron 3 Diarization tells up to 8 voices apart, on your computer. Rename speakers once you know who they are, merge two that are the same person, or move a single line to someone else.

Online converters upload your file to a server and count your minutes. Glimpse transcribes on your computer, so a confidential interview or a patient recording never leaves it, and a three hour lecture costs the same as a three minute voice memo. It works with the wifi off once the model is downloaded.

Copy the transcript, or export it as plain text, Markdown, SRT or VTT subtitles. Everything stays in your Library, searchable alongside your recordings and dictations.

Pick the model per job. Whisper covers 99 languages and detects which one is spoken. Parakeet covers 25 European languages and is the fastest. Your custom dictionary helps with names and jargon in either.