Skip to main content

Voice Translation Settings

Here you can configure the feature switches for voice translation, the microphone and speaker devices, the translation languages, speech synthesis (TTS), the translation result display, and VRChat Chatbox sending. Configurations that are strongly tied to the model, such as the speech recognition model and the inference device, have been moved to the "ASR Settings / TTS Settings" cards of Translation Engine Settings. For functional usage instructions, please refer to Voice Translation.

Feature Settings​

  • Enable Voice Translation: The master switch for voice translation. When enabled, the main program will start the voice worker process and begin capturing audio.
  • Enable Speaker: Whether to enable the speaker (loopback capture) channel.
  • Enable Microphone: Whether to enable the microphone channel.
  • Enable Speech Synthesis (TTS): Whether to synthesize the translation of the microphone channel into speech and read it aloud.

Configurations that are strongly tied to the model, such as the speech recognition model, inference device, inference core count, and simulated streaming output, have been moved to the "ASR Settings" card of Translation Engine Settings; TTS streaming playback, inference device, voice, and voice cloning have been moved to the "TTS Settings" card on that page.

Speaker Voice Translation Settings​

Configure for speaker loopback capture (i.e., what others say during games or voice calls):

  • Speech Recognition Language: Set the language used for recognizing speaker audio, default is "Auto", letting the system determine the language automatically.
  • Translation Target Language: Set the translation target language for the recognized speaker text. The target language is usually set to your native language, to understand what the other person is saying.

The translation source language is fixed to "Auto", and the translation engine automatically determines the language based on the recognized text; the settings no longer provide an entry point for selecting the source language.

Microphone Voice Translation Settings​

Configure for the microphone (i.e., your own speech), used to translate your words for others:

  • Speech Recognition Language: Default is "Auto".
  • Translation Target Language: Set the translation target language of your speech, for example target language English, making it convenient to communicate with foreign players in multilingual social settings.

Audio Device Settings​

  • Capture Source Note: If some headphone devices cannot capture system audio, please select the corresponding program for window capture.
  • Select Application: The capture source for the speaker. Selecting "System" means capturing the entire system; you can also specify a specific application to capture only that program's audio. Saved programs remain in the options even if they are not currently running.
  • Microphone Device: Select the input device that captures your voice.
  • Volume Threshold (Microphone): The red threshold line on the microphone volume bar; you can drag or click to position it, and the "↺" button on the right restores the default. When the volume crosses the threshold, the volume bar fill turns green, indicating that it is judged as valid speech.
  • Speaker Device: Select the device used for loopback capture of system audio. When the capture source is specified as an application, this item is automatically grayed out (in that case, the entire system is no longer captured by device).
  • Volume Threshold (Speaker): The threshold of the speaker volume bar is adjusted in the same way as for the microphone.

The device list is automatically enumerated and periodically refreshed by the subprocess, so there is no need to refresh it manually. If the volume bar does not respond, please confirm that voice translation is enabled—real-time volume can only be monitored after voice translation is enabled.

Speech Synthesis (TTS)​

Synthesize the translation of the microphone channel into speech and read it aloud. Qwen3-TTS runs in the voice worker process, and the model loads on demand (loaded into memory only on the first read-aloud request) and automatically unloads after 30 seconds of inactivity.

  • Speaker Synchronized Playback: When enabled, the TTS audio is also sent to the speaker device in the audio settings; during synchronized playback, recognition on the speaker channel is automatically suppressed to prevent your own translation from being recognized back.
  • Route the Microphone Device Directly to the TTS Playback Device When TTS Is Disabled: When read-aloud is unavailable (TTS off / voice translation off / paused), the microphone capture is forwarded directly to the TTS playback device, so the in-game microphone does not need to switch devices frequently.
  • Device Mapping Tip: Displays the tip "For the playback output device, select CABLE Input; for the in-game microphone device, select CABLE Output".
  • Playback Output Device: The output device for read-aloud audio. When the virtual audio cable's CABLE Input is detected, it is automatically selected as the default (only when the configuration is empty).
  • Playback Volume: The read-aloud volume (0.0 to 2.0, default 1.0, where 1.0 is the original level).

This feature requires installing the third-party virtual audio driver VB-Audio Virtual Cable. When CABLE Input is not detected, the interface will display a prompt along with the official website address.

For settings such as the voice, voice cloning, TTS real-time streaming playback, and TTS inference device, please go to the "TTS Settings" card of Translation Engine Settings.

Translation Result Display Settings​

  • Display Mode: Comparison mode (shows both original and translation) or translation-only mode (shows only the translation).
  • Translation Result Alignment: The left / center / right alignment of the text within the translation panel.
  • Translation Result Flow Direction: The scrolling direction of newly added content, selectable as Auto / Up / Down.
  • Background Transparency: The transparency of the translation panel background (0.0 to 1.0, default 0.5).
  • Font Size: The scaling ratio of the translation text (0.5x to 2.0x, default 1.0x).
  • Width / Height: The proportion of the translation panel in the field of view, adjustable between 10% ~ 60% (width) and 5% ~ 30% (height).
  • X Position / Y Position: The display position of the translation panel center (0% ~ 100%).
  • Depth (m): The display depth distance of the translation panel in VR space (0.5 to 10.0 meters, default 2.0).
  • Translation Result Display Method: Follow (the translation panel follows the line of sight) or Fixed (pinned to a position in VR space).

VRChat Chatbox (OSC)​

Automatically send voice translation results to VRChat's Chatbox via the OSC protocol:

  • Send Speaker Translation Text to Chatbox: Whether to send the translation of the speaker (what the other person says).
  • Send Microphone Translation Text to Chatbox: Whether to send the translation of the microphone (what you say).
  • OSC Port: The VRChat receiving port, default is 9000.
  • OSC Message Mode: Comparison mode (translation on top, original below) or translation-only mode.
  • Message Flow Direction: New messages are displayed in the Down or Up direction.
  • Number of Retained Messages: The number of history messages retained in the chat box (1 to 10, default 2).

Tip: Voice translation requires the voice worker process to run properly. If this feature cannot be enabled or behaves abnormally, please confirm that the software installation directory is intact, and check whether the antivirus software is blocking the subprocess.