AI-powered audio transcription using OpenAI's remarkably accurate Whisper speech recognition system, with support for numerous languages, as well as translation #Audio Transcription #OpenAI Whisper #AI Transcription #Transcription #Whisper #OpenAI
OpenAI has recently unveiled Whisper, a speech recognition system that can be used to accurately transcribe audio. It relies on a massive and diverse dataset, with the aim of ensuring more reliable recognition of natural speech. It supports dozens of languages and is even capable of on-the-fly translation.
Aiko is a Mac app that provides an intuitive, accessible GUI for using Whisper on your desktop. It can transcribe imported audio or video files, and you can also record your voice and have the app process it directly.
Whisper is truly impressive. It is able to accurately detect language even from less-than-ideal sources, and its ability to apply correct punctuation and organize text into paragraphs makes you wonder just how good AI can get.
After feeding it both English and French audio, recordings that were not exactly recent or high-quality, I found virtually no errors. Punctuation was sometimes different from human transcriptions, but not in a way that was grammatically incorrect.
It can also take audio in any supported language and translate it into English as it is being transcribed. The potential uses for this kind of technology are almost endless.
Aiko makes it very easy to take advantage of OpenAI’s system. You can load audio or video files and wait for the app to transcribe them, or simply start recording your voice using a microphone. The language is detected automatically, but you can also select it yourself.
Once the transcription is completed, you can copy the text to the clipboard or export it. Editing is not supported, and you can’t add to a completed transcription. If you want to record something else, you have to start over.
There are a couple of small issues that could be addressed. First off, while automatic translation can be disabled, we encountered a situation where it was difficult to avoid. If a file’s audio was mostly in a different language, but it started in English, all non-English text would also be translated regardless of the app’s settings.
Secondly, transcribing microphone audio could be improved. There is no way to pause a recording, and if you stop the transcription to take a break, you can’t pick up where you left off. You have to save the text you’ve already transcribed and start a new recording. Also, the ability to preview transcribed text would help.
Overall, Aiko is a great way to access the capabilities of OpenAI’s Whisper on your Mac. It’s not a prefect app, but it’s well designed. It also works entirely offline and is completely free, since it doesn’t need to rely on the Whisper API — the entire model is included and accessed from your own storage.
What's new in Aiko 1.7.0:
- Added setting for word replacements.
Aiko 1.7.0
add to watchlist add to download basket send us an update REPORT- runs on:
- macOS 14.4 or later (Universal Binary)
- file size:
- 2.9 GB
- main category:
- Audio
- developer:
- visit homepage
ShareX
4k Video Downloader
Bitdefender Antivirus Free
Windows Sandbox Launcher
Context Menu Manager
Zoom Client
IrfanView
calibre
7-Zip
Microsoft Teams
- calibre
- 7-Zip
- Microsoft Teams
- ShareX
- 4k Video Downloader
- Bitdefender Antivirus Free
- Windows Sandbox Launcher
- Context Menu Manager
- Zoom Client
- IrfanView