Tom: Transcribe Audio & Video

Transcribe audio & video with the engine that fits the job — three options, all 100% on-device.

Tom turns sound into clean, editable text right on your iPhone. Lectures, WhatsApp voice notes, screen recordings, interviews, podcasts, long meetings — pick your engine, drop in the file, read the transcript. No account, no upload, no subscription.

Built originally for a friend who is deaf, Tom is for anyone who needs to read what was said.

CHOOSE YOUR ENGINE

• Apple Speech — fast, low memory, built into iOS. Great default for live recording and short files. 25 languages.
• Whisper — OpenAI's recogniser, running locally. Six sizes from Tiny (80 MB) to Large v3 (3.1 GB), including Turbo for the best speed/quality balance. Up to 99 languages.
• Parakeet v3 — high-accuracy European transcription from FluidAudio, on-device. 25 European languages. Apple Silicon, iOS 17+.

Not happy with the result? Re-transcribe the same file with a different engine in one tap — the previous version is kept in the version history, so you can compare and switch back.

KEY FEATURES

• Live recording — tap once and watch the transcript appear as you speak.
• File import — any video from your library, or audio files from Files (including WhatsApp voice notes saved to Files).
• Multiple engines + re-transcribe — pick Apple Speech, Whisper, or Parakeet per file, and re-run the same audio through a different engine without losing the previous transcript.
• Editable segments — long-press any segment to fix a word the recogniser got wrong. Your edit stays on the device.
• Subtitle export — save any transcript as plain text, SubRip (.srt) or WebVTT (.vtt) for accessible video.
• Player with smart controls — tap-to-seek, follow the highlighted segment as it plays, skip ±10 seconds, and switch speed to 1.5x or 2x.
• Honest hallucination flag — when Whisper invents text on silence or music, suspect segments are clearly marked so you can verify or remove them. We never silently strip your transcript.
• Long files — chunked processing for meetings, lectures, conferences and full episodes.
• Model management — downloaded Whisper and Parakeet models live in Documents, visible in the Files app. Delete a model from Settings or from Files to reclaim space anytime.
• Wi-Fi only downloads — optional, on by default, so large models never eat your mobile data.
• Smart storage — save transcriptions with custom titles, search by title or content.
• Activity log for transparency on every step.

WHO IS IT FOR

• Students extracting notes from long lectures, videos, or screen recordings.
• Professionals transcribing interviews, meetings, or lengthy audio files.
• WhatsApp users who want to read voice messages instead of listening.
• Content creators turning audio or video into text or subtitles for TikTok, Reels, Shorts, or blog posts.
• Podcasters making episodes accessible with .srt subtitles.
• Anyone who is deaf or hard of hearing and needs reliable, private transcription.

WHY CHOOSE TOM

Mos