Introduction

The core philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When building agents that need to handle voice interaction, a common requirement is to recognize regional dialects and accents with low cost and high accuracy. The xby-asr-f plugin is designed for this purpose. It provides a suite of speech recognition interfaces based on the Xiaobenyang service, supporting Mandarin Chinese and more than 20 other dialects and accents.

Plugin Overview

This is a model inference plugin maintained by xby-skill and released under the MIT License. Its core value lies in encapsulating complex dialect speech recognition capabilities as standard DSH interfaces. Developers do not need to worry about the underlying recognition logic; they only need to call the interfaces according to the specification to convert speech to text.

Installation

Before installation, ensure that your DeepSeek Harness environment is properly configured. Install the plugin via the command line:

dsh plugin --profile web add github:xby_skill/xby-asr-f

After installation is complete, the plugin will be loaded when the DSH process starts.

API Key Configuration

Before using this plugin, a valid API key must be configured. You can set the key by directly invoking the set_xby_apikey command in a chat conversation. This command supports automatic persistence, so the configuration remains valid after restarting DSH.

Speech Recognition Usage

The plugin provides three recognition methods. Choose the appropriate method based on the input data type:

  1. asr_fire_red_asr_ctc: accepts a URL link to an audio file.
  2. asr_fire_red_asr_ctc_for_data_base64: accepts the BASE64 encoding of an audio file.
  3. asr_fire_red_asr_ctc_for_data_file: accepts a local file path to an audio file.

Notes

Please note that the recognition service depends on the official Xiaobenyang website. Before use, you need to visit the official Xiaobenyang website to register and obtain an API key. Additionally, the plugin runs with the permissions of the current dsh process. It is recommended to review the source code and license before installation.

Conclusion

In summary, xby-asr-f adds dialect speech recognition capabilities to the DSH ecosystem. If you need to build agents that handle voice input from multiple regions, you can get more information or submit feedback through the following links:
* GitHub repository
* DSH plugin directory