flat: 优化语音

hook/useRealtimeRecorder.js

@@ -31,8 +31,8 @@ export function useAudioRecorder(wsUrl) {
     // 配置常量
     const SAMPLE_RATE = 16000;
     const SILENCE_THRESHOLD = 0.02; // 静音阈值 (0-1)
-    const SILENCE_DURATION = 400; // 静音持续时间(ms)后切片
-    const MIN_SOUND_DURATION = 300; // 最小有效声音持续时间(ms)
+    const SILENCE_DURATION = 100; // 静音持续时间(ms)后切片
+    const MIN_SOUND_DURATION = 200; // 最小有效声音持续时间(ms)
 
     // 音频处理变量
     const lastSoundTime = ref(0);
@@ -125,6 +125,7 @@ export function useAudioRecorder(wsUrl) {
             socket.value = new WebSocket(wsUrl);
 
             socket.value.onopen = () => {
+                console.log('open')
                 isSocketConnected.value = true;
                 resolve();
             };
@@ -191,7 +192,6 @@ export function useAudioRecorder(wsUrl) {
         if (audioChunks.value.length === 0 || !socket.value || socket.value.readyState !== WebSocket.OPEN) {
             return;
         }
-
         try {
             // 合并所有块
             const totalBytes = audioChunks.value.reduce((total, chunk) => total + chunk.byteLength, 0);

hook/useTTSPlayer.js

@@ -0,0 +1,249 @@
+import {
+    ref,
+    onUnmounted,
+    onBeforeUnmount,
+    onMounted
+} from 'vue'
+import {
+    onHide,
+    onUnload
+} from '@dcloudio/uni-app'
+import WavDecoder from '@/lib/wav-decoder@1.3.0.js'
+
+export function useTTSPlayer(wsUrl) {
+    const isSpeaking = ref(false)
+    const isPaused = ref(false)
+    const isComplete = ref(false)
+
+    const audioContext = new(window.AudioContext || window.webkitAudioContext)()
+    let playTime = audioContext.currentTime
+    let sourceNodes = []
+    let socket = null
+    let sampleRate = 16000
+    let numChannels = 1
+    let isHeaderDecoded = false
+    let pendingText = null
+
+    let currentPlayId = 0
+    let activePlayId = 0
+
+    const speak = (text) => {
+        currentPlayId++
+        const myPlayId = currentPlayId
+        reset()
+        pendingText = text
+        activePlayId = myPlayId
+    }
+
+    const pause = () => {
+        if (audioContext.state === 'running') {
+            audioContext.suspend()
+            isPaused.value = true
+            isSpeaking.value = false
+        }
+    }
+
+    const resume = () => {
+        if (audioContext.state === 'suspended') {
+            audioContext.resume()
+            isPaused.value = false
+            isSpeaking.value = true
+        }
+    }
+
+    const cancelAudio = () => {
+        stop()
+    }
+
+    const stop = () => {
+        isSpeaking.value = false
+        isPaused.value = false
+        isComplete.value = false
+        playTime = audioContext.currentTime
+
+        sourceNodes.forEach(node => {
+            try {
+                node.stop()
+                node.disconnect()
+            } catch (e) {}
+        })
+        sourceNodes = []
+
+        if (socket) {
+            socket.close()
+            socket = null
+        }
+
+        isHeaderDecoded = false
+        pendingText = null
+    }
+
+    const reset = () => {
+        stop()
+        isSpeaking.value = false
+        isPaused.value = false
+        isComplete.value = false
+        playTime = audioContext.currentTime
+        initWebSocket()
+    }
+
+    const initWebSocket = () => {
+        const thisPlayId = currentPlayId
+        socket = new WebSocket(wsUrl)
+        socket.binaryType = 'arraybuffer'
+
+        socket.onopen = () => {
+            if (pendingText && thisPlayId === activePlayId) {
+                const seepdText = extractSpeechText(pendingText)
+                socket.send(seepdText)
+                pendingText = null
+            }
+        }
+
+        socket.onmessage = async (e) => {
+            if (thisPlayId !== activePlayId) return // 忽略旧播放的消息
+
+            if (typeof e.data === 'string') {
+                try {
+                    const msg = JSON.parse(e.data)
+                    if (msg.status === 'complete') {
+                        isComplete.value = true
+                        setTimeout(() => {
+                            if (thisPlayId === activePlayId) {
+                                isSpeaking.value = false
+                            }
+                        }, (playTime - audioContext.currentTime) * 1000)
+                    }
+                } catch (e) {
+                    console.log('[TTSPlayer] 文本消息:', e.data)
+                }
+            } else if (e.data instanceof ArrayBuffer) {
+                if (!isHeaderDecoded) {
+                    try {
+                        const decoded = await WavDecoder.decode(e.data)
+                        sampleRate = decoded.sampleRate
+                        numChannels = decoded.channelData.length
+                        decoded.channelData.forEach((channel, i) => {
+                            const audioBuffer = audioContext.createBuffer(1, channel.length,
+                                sampleRate)
+                            audioBuffer.copyToChannel(channel, 0)
+                            playBuffer(audioBuffer)
+                        })
+                        isHeaderDecoded = true
+                    } catch (err) {
+                        console.error('WAV 解码失败:', err)
+                    }
+                } else {
+                    const pcm = new Int16Array(e.data)
+                    const audioBuffer = pcmToAudioBuffer(pcm, sampleRate, numChannels)
+                    playBuffer(audioBuffer)
+                }
+            }
+        }
+    }
+
+    const pcmToAudioBuffer = (pcm, sampleRate, numChannels) => {
+        const length = pcm.length / numChannels
+        const audioBuffer = audioContext.createBuffer(numChannels, length, sampleRate)
+        for (let ch = 0; ch < numChannels; ch++) {
+            const channelData = audioBuffer.getChannelData(ch)
+            for (let i = 0; i < length; i++) {
+                const sample = pcm[i * numChannels + ch]
+                channelData[i] = sample / 32768
+            }
+        }
+        return audioBuffer
+    }
+
+    const playBuffer = (audioBuffer) => {
+        if (!isSpeaking.value) {
+            playTime = audioContext.currentTime
+        }
+        const source = audioContext.createBufferSource()
+        source.buffer = audioBuffer
+        source.connect(audioContext.destination)
+        source.start(playTime)
+        sourceNodes.push(source)
+        playTime += audioBuffer.duration
+        isSpeaking.value = true
+    }
+
+    onUnmounted(() => {
+        stop()
+    })
+
+    // 页面刷新/关闭时
+    onMounted(() => {
+        if (typeof window !== 'undefined') {
+            window.addEventListener('beforeunload', cancelAudio)
+        }
+    })
+
+    onBeforeUnmount(() => {
+        cancelAudio()
+        if (typeof window !== 'undefined') {
+            window.removeEventListener('beforeunload', cancelAudio)
+        }
+    })
+
+    onHide(cancelAudio)
+    onUnload(cancelAudio)
+
+    initWebSocket()
+
+    return {
+        speak,
+        pause,
+        resume,
+        cancelAudio,
+        isSpeaking,
+        isPaused,
+        isComplete
+    }
+}
+
+function extractSpeechText(markdown) {
+    const jobRegex = /``` job-json\s*({[\s\S]*?})\s*```/g;
+    const jobs = [];
+    let match;
+    let lastJobEndIndex = 0;
+    let firstJobStartIndex = -1;
+
+    // 提取岗位 json 数据及前后位置
+    while ((match = jobRegex.exec(markdown)) !== null) {
+        const jobStr = match[1];
+        try {
+            const job = JSON.parse(jobStr);
+            jobs.push(job);
+            if (firstJobStartIndex === -1) {
+                firstJobStartIndex = match.index;
+            }
+            lastJobEndIndex = jobRegex.lastIndex;
+        } catch (e) {
+            console.warn('JSON 解析失败', e);
+        }
+    }
+
+    // 提取引导语（第一个 job-json 之前的文字）
+    const guideText = firstJobStartIndex > 0 ?
+        markdown.slice(0, firstJobStartIndex).trim() :
+        '';
+
+    // 提取结束语（最后一个 job-json 之后的文字）
+    const endingText = lastJobEndIndex < markdown.length ?
+        markdown.slice(lastJobEndIndex).trim() :
+        '';
+
+    // 岗位信息格式化为语音文本
+    const jobTexts = jobs.map((job, index) => {
+        return `第 ${index + 1} 个岗位，岗位名称是：${job.jobTitle}，公司是：${job.companyName}，薪资：${job.salary}，地点：${job.location}，学历要求：${job.education}，经验要求：${job.experience}。`;
+    });
+
+    // 拼接总语音内容
+    const finalTextParts = [];
+    if (guideText) finalTextParts.push(guideText);
+    finalTextParts.push(...jobTexts);
+    if (endingText) finalTextParts.push(endingText);
+
+    return finalTextParts.join('\n');
+}