Doubao AST 2.0 Setup Tutorial

1Register a Volcengine Account

Go to volcengine.com and sign up for an account. If you already have a ByteDance/Volcengine account, you can sign in directly.

2Navigate to the Speech Console

After logging in, select "Doubao Simultaneous Interpreting 2.0" from the left menu, or visit the Speech Console directly.

3Create an Application

In the Speech console, click "Create Application" to create a new application. Name it (e.g., "Sokuji Translation") and confirm creation.

4Get APP ID and Access Token

Click "Doubao Simultaneous Interpreting 2.0" in the left menu. Select your newly created application from the application name dropdown at the top of the page, then copy the APP ID and Access Token shown below. Sokuji also accepts a Volcengine API Key; either kind of credential will do.

5Configure in Sokuji

Open Sokuji and select "Doubao AST 2.0" as your AI provider in the settings panel. Choose the credential kind App ID + Access Token and enter your APP ID and Access Token, or choose API key and paste the API Key. Then choose the source and target languages; Speech Mode (Auto, Push-to-Talk or Push-to-Translate) is one setting for every provider.

6Test the Connection

Click the Start Session button and speak into your microphone to verify that speech-to-speech translation is working. You should see the translation. The spoken translation goes into the virtual microphone for the other side; to hear it yourself, turn on I hear it too in the Speech block (Quick settings, or Settings → Advanced → Audio).

What is Doubao AST 2.0?

Doubao AST 2.0 is an end-to-end speech translation service:

• Speech-to-speech simultaneous translation without intermediate text steps
• Voice cloning: translated speech preserves the speaker's voice characteristics
• Fixed voices: pick one per target language from Doubao's library, filtered by gender, age and style, with a preview
• Voice I hear: with Translation I hear on in the Speech block, the other side's translation is read aloud to you too, and Voice Settings shows two rows, Voice the other side hears and Voice I hear. Voice I hear lists the voices of your own language; Clone the speaker's voice there clones the other person. A voice is kept per language, so a voice you pick for a language speaks it in either row. With Chinese + English, one row serves both sides
• Low-latency real-time translation processing
• Chinese and English can be translated both ways in one session

Supported Languages

One side of every language pair must be Chinese or English:

• Spoken (speech-to-speech): Chinese, English, Japanese, Indonesian, Spanish, Portuguese, German, French; from Chinese or English, Korean can be spoken too
• Text only: Korean, Turkish, Malay, Dutch, Romanian, Polish, Czech, Arabic, Thai, Vietnamese, Russian, Italian
• Dialects: Cantonese and Wu (Shanghainese), as the source language only
• “Chinese + English”: both languages translated each way in one session

Speech Mode

• Auto: the service detects when speech starts and ends, suitable for natural conversations
• Push-to-Talk: audio is sent only while you hold Space or the microphone button, suitable for noisy environments or when precise control is needed
• Push-to-Translate: the same as Push-to-Talk, except that while you are not holding, your own voice goes straight into the virtual microphone

Common Issues

Authentication Error: Verify that your APP ID and Access Token (or API Key) are correctly copied without extra spaces, and that the credential kind selected in Sokuji matches what you entered.

Connection Error: Check your network connection and ensure firewalls are not blocking WebSocket connections. If using a proxy, make sure it supports WebSocket.

Unsupported Language Pair: Confirm that your selected source and target language combination is within the supported range of Doubao AST 2.0.

Tips for Best Results

1. Use a stable network connection for best translation quality
2. Use in a quiet environment for the best speech recognition
3. If translation quality is not ideal, try switching the Speech Mode
4. Regularly check usage and quota in the Volcengine console