同声传译 WebSocket

1. 客户端建立 WebSocket 连接
   ↓
2. 服务端返回 connected_success 事件
   ↓
3. 客户端发送 task_start 事件
   ↓
4. 服务端返回 task_started 事件
   ↓
5. 客户端持续发送二进制 PCM 音频帧
   ↓
6. 服务端返回原文、译文、译文 TTS 事件
   ↓
7. 客户端发送 task_finish 事件
   ↓
8. 服务端返回 task_finished 事件并关闭连接

{
  "event": "task_start",
  "model": "sensenova-livetranslate-1.0",
  "audio_setting": {
    "sample_rate": 16000,
    "format": "pcm",
    "channel": 1
  },
  "voice_input_setting": {
    "language": "Chinese",
    "target_language": "English"
  },
  "tts_setting": {
    "enable": false
  }
}

{
  "event": "task_finish"
}

{
  "event": "connected_success",
  "session_id": "session-id-xxx",
  "base_resp": {
    "status_code": 0,
    "status_msg": "success"
  }
}

pip install websockets

import asyncio
import json
import os

import websockets

WS_URL = "wss://api.senseaudio.cn/ws/v1/audio/simulat-interpreting"
API_KEY = os.environ["SENSEAUDIO_API_KEY"]
AUDIO_FILE = "input_16k_s16le_mono.pcm"


async def wait_for_event(ws, expected):
    while True:
        message = json.loads(await ws.recv())
        if message.get("event") == "task_failed":
            raise RuntimeError((message.get("base_resp") or {}).get("status_msg"))
        if message.get("event") == expected:
            return message


async def receive_events(ws):
    async for raw in ws:
        message = json.loads(raw)
        event = message.get("event")
        data = message.get("data") or {}

        if event == "si_segment_done":
            print(f"原文: {data.get('original_text', '')}")
            print(f"译文: {data.get('text', '')}")
        elif event == "task_failed":
            raise RuntimeError((message.get("base_resp") or {}).get("status_msg"))
        elif event == "task_finished":
            return


async def main():
    headers = {"Authorization": f"Bearer {API_KEY}"}

    async with websockets.connect(WS_URL, additional_headers=headers) as ws:
        await wait_for_event(ws, "connected_success")

        await ws.send(json.dumps({
            "event": "task_start",
            "model": "sensenova-livetranslate-1.0",
            "audio_setting": {
                "sample_rate": 16000,
                "format": "pcm",
                "channel": 1,
            },
            "voice_input_setting": {
                "language": "Chinese",
                "target_language": "English",
            },
            "tts_setting": {"enable": False},
        }))
        await wait_for_event(ws, "task_started")

        receive_task = asyncio.create_task(receive_events(ws))

        with open(AUDIO_FILE, "rb") as audio:
            while chunk := audio.read(3200):
                await ws.send(chunk)
                await asyncio.sleep(0.1)

        await ws.send(json.dumps({"event": "task_finish"}))
        await receive_task


asyncio.run(main())

npm install ws

const fs = require('fs')
const WebSocket = require('ws')

const WS_URL = 'wss://api.senseaudio.cn/ws/v1/audio/simulat-interpreting'
const API_KEY = process.env.SENSEAUDIO_API_KEY
const AUDIO_FILE = 'input_16k_s16le_mono.pcm'

const ws = new WebSocket(WS_URL, {
  headers: {
    Authorization: `Bearer ${API_KEY}`,
  },
})

let audioStream = null

ws.on('message', (raw, isBinary) => {
  if (isBinary) return

  const message = JSON.parse(raw.toString())
  const event = message.event
  const data = message.data || {}

  switch (event) {
    case 'connected_success':
      ws.send(JSON.stringify({
        event: 'task_start',
        model: 'sensenova-livetranslate-1.0',
        audio_setting: {
          sample_rate: 16000,
          format: 'pcm',
          channel: 1,
        },
        voice_input_setting: {
          language: 'Chinese',
          target_language: 'English',
        },
        tts_setting: { enable: false },
      }))
      break

    case 'task_started':
      audioStream = fs.createReadStream(AUDIO_FILE, { highWaterMark: 3200 })
      audioStream.on('data', (chunk) => {
        audioStream.pause()
        ws.send(chunk, { binary: true }, () => {
          setTimeout(() => audioStream.resume(), 100)
        })
      })
      audioStream.on('end', () => {
        ws.send(JSON.stringify({ event: 'task_finish' }))
      })
      break

    case 'si_segment_done':
      console.log('原文:', data.original_text || '')
      console.log('译文:', data.text || '')
      break

    case 'task_failed':
      console.error('任务失败:', message.base_resp?.status_msg)
      break
  }
})