Aiko — Sindre Sorhus
Simplifying Audio Transcription with Aiko
Transcribing audio to text has been made easier with the introduction of Aiko, a tool that utilizes artificial intelligence (AI) to provide high-quality transcriptions directly on your device.
High-Quality Transcription On-Device
Aiko uses the Whisper model from OpenAI to transcribe meetings, lectures, and voice notes with ease. The app runs locally on your device, ensuring privacy for sensitive recordings. It also supports Shortcuts on iOS for a streamlined experience.
Upcoming Enhancements
Aiko promises new features such as batch conversation capabilities, exporting transcriptions to a karaoke file format, and support for audio in 100 languages.
Maintaining Your Privacy
Aiko processes transcriptions on your device without sending data to external servers, prioritizing privacy and confidentiality.
Technical Specifications
Aiko uses the large v2 Whisper model on macOS and the medium or small models on iOS devices, depending on available memory.
Helpful Transcription Tips
Aiko divides text by sentences, but users can format it into paragraphs using GPT-4 or GPT-3.5 prompts. It also provides a solution for missing punctuation.
Addressing Common Concerns
Aiko's Mac app uses the v2 model because it performs better. Editing is not supported within the app, but text can be exported and edited using any text editor. Aiko offers superior accuracy and language support compared to Apple's built-in transcription services. Any errors in the transcription are not within the developer's control, but feedback is welcome. Aiko may sometimes repeat phrases or miss punctuation due to limitations in the AI model. Extra sentences in the transcription are unintentional quirks from the AI's training.
Summary
Aiko simplifies audio transcription with its modern, efficient, and private approach. While it has some flaws related to the AI model, its convenience, security, and language support make it an attractive option for converting speech to text.
For more information or support, users can refer to the app's FAQs.
