xby-asr-1

Speech-to-text in DeepSeek Harness conversations has traditionally meant leaving the chat, opening a separate tool, and pasting the result back. xby-asr-1 pulls that step into the conversation itself: hand the plugin an audio source and the transcript appears right where you are. It addresses a practical pain point—making transcription a single in-chat operation rather than a multi-tool workflow.

Built on the Whisper Small model, the plugin handles recognition across many languages and minority languages alike. What sets it apart is input flexibility: you can supply a URL to an audio file, a Base64-encoded payload, or a local file path. Three entry points cover the most common ways audio material is stored, so there is no forced conversion step. The API key is persisted automatically after a single configuration, surviving restarts and saving you from re-entering credentials every session.

This plugin is a good fit for developers who process meeting recordings, podcast clips, or multilingual voice messages frequently within a conversation flow, and for anyone who wants to quickly turn spoken content into editable text for further analysis. It does not change how you work; it simply removes one detour from the speech-to-text path.