Volcengine Doubao Speech

Redock can use Volcengine Doubao speech recognition for Chinese and mixed Chinese-English terminal input.

What Redock Calls

Redock sends audio to:

POST https://openspeech.bytedance.com/api/v3/auc/bigmodel/recognize/flash

The request uses:

• Auth header: X-Api-Key
• Resource header: X-Api-Resource-Id
• Request body model name: bigmodel
• Audio upload: Base64 audio data

Redock records 16 kHz mono WAV audio before sending it.

Enable the Service

1. Sign in to the Volcengine console.
2. Open the Doubao Speech console.
3. Enable the speech recognition large model service.
4. Confirm the fast recorded-file recognition API is available.
5. Confirm the account or project can use the Resource ID you plan to enter.

https://console.volcengine.com/speech/app

The default Resource ID in the official fast recognition API documentation is:

volc.bigasr.auc_turbo

Get the API Key

1. Open API Key management in the Doubao Speech console.
2. Create a new API Key or copy an existing valid one.
3. Keep it private.

Redock stores the key in Keychain. It is not written into app data snapshots.

Fill It in Redock

1. Open My > Speech.
2. Set Speech Provider to Volcengine Doubao.
3. Set Resource ID. Start with volc.bigasr.auc_turbo unless your console shows a different compatible resource.
4. Paste the APP Key.
5. Return to a terminal workspace and test speech input.

Common Issues

• Missing API key: check My > Speech.
• Permission or auth failed: confirm the account has enabled the Resource ID.
• Empty result: try louder or longer audio.
• Wrong field: ep-... endpoint IDs for text inference are not the Resource ID needed here.
• Billing or trial issue: check service status, quota, and billing in the Volcengine console.

Official References

• Doubao Speech console
• Fast recorded-file recognition API
• Doubao Speech API Key usage