Aiko vs noScribe
Aiko
sindresorhus.com
noScribe
noscribe.deSide by side
| Figure | Aiko | noScribe |
|---|---|---|
| Price from | — | Free |
| Free plan | — | Yes |
| Free trial | Yes | — |
| Open source | — | — |
| Windows | — | Yes |
| macOS | Yes | Yes |
| Linux | — | Yes |
| iOS | Yes | — |
| Plans published | 1 | 1 |
| Spec score · Transcription Software | 8.6 (#12) | 8.4 (#20) |
| Spec score · AI transcription tools | 8.1 (#26) | 8.3 (#22) |
| File support | Aiko accepts audio and video formats supported by macOS and iOS, including M4A, WAV, MP3, MP4, and MOV, but not OGG. | The app can process almost any audio or video format. |
| Free plan | No | Yes |
| Privacy | Aiko transcribes audio directly on the device, and its privacy policy says the app collects no data or personal information. | Transcription runs entirely on the user's computer, without cloud processing. |
| Purpose | Aiko converts speech to text from meetings, lectures, and other audio. | noScribe is a desktop app for transcribing interviews and other sensitive audio data with local AI models. |
| Speaker identification | No | Yes |
| Support | Users can submit feedback or bug reports through the app or the maker’s feedback form; the form says to check spam if no reply arrives within two weeks. | The maker recommends GitHub Issues for help and provides email contact for urgent cases. |
Plans
In their own words
noScribe
noScribe is a free, open-source desktop app for transcribing interviews and sensitive recordings with AI models running locally on the computer. It uses Whisper-based models for transcription and pyannote for speaker recognition, with support for around 60 languages. Its editor aligns text with audio playback so users can review and correct transcripts. Research tools let users mark pauses and overlapping speech, add timestamps, and choose whether filler words are included. The app accepts almost any audio or video format and exports formatted HTML, plain text, or WebVTT subtitles. Transcripts can be opened in MS Word or LibreOffice, or imported into QDA software such as MAXQDA or QualCoder. The maker estimates a one-hour interview takes about one to three hours to transcribe; NVIDIA CUDA acceleration can reduce processing time. Current versions support Windows 10 and require macOS 14 or later. Linux binaries are not available for version 0.8; users are directed to version 0.7 or manual installation. It is licensed under GPL-3.0.