To realize speech to text in offline environment, we recommend using Spokenly's native Whisper modeling feature. The steps are as follows:
- In Spokenly's settings, select 'Speech Model' and switch to the 'Local Whisper Model' option.
- Make sure the device meets the minimum configuration requirements (macOS 12.0 or higher, 8GB RAM recommended).
- To improve recognition accuracy, use a high-quality microphone in a quiet environment and use articulation clarity as a primary criterion.
- For special terms or specialized vocabulary, replacement rules can be pre-set in the AI prompts (e.g. 'replace CRM with Customer Relationship Management System').
Note that the local model has weak punctuation processing, which can be added manually at a later stage or by setting up AI prompting rules (e.g., 'convert periods to.') to optimize the output results.
This answer comes from the articleSpokenly: a speech-to-text tool for macOSThe
May not be reproduced without permission:AI productivity tools " How to Achieve Highly Accurate Speech to Text in a Completely Offline Environment
English 
简体中文 
日本語 
Deutsch 
Português do Brasil