node-red-node-rdk-tools/src/rdk-speechtotext/rdk-speechtotext.html

配合RDK硬件及TROS使用的Node-RED功能包(Node-RED nodes for using TROS on a RDK hardware and TROS)

<script type="text/x-red" data-template-name="rdk-tools speechtotext">
    <div class="form-row node-input-name">
        <label for="node-input-name"><i class="fa fa-tag"></i> <span data-i18n="rdk-speechtotext.label.name"></span></label>
        <input type="text" id="node-input-name" data-i18n="[placeholder]rdk-speechtotext.names.speechtotext" style="width: 296px;">
    </div>
</script>

<script type="text/javascript">
(function() {
    function encodeWAV(audioBuffer) {
        const numChannels = audioBuffer.numberOfChannels;
        const sampleRate = audioBuffer.sampleRate;
        const format = 1; // PCM
        const bitsPerSample = 16;
        const samples = audioBuffer.length;
        const blockAlign = numChannels * bitsPerSample / 8;
        const byteRate = sampleRate * blockAlign;
        const dataSize = samples * blockAlign;
        const buffer = new ArrayBuffer(44 + dataSize);
        const view = new DataView(buffer);

        // Write WAV header
        let offset = 0;
        function writeString(str) {
            for (let i = 0; i < str.length; i++) {
            view.setUint8(offset++, str.charCodeAt(i));
            }
        }

        writeString("RIFF");
        view.setUint32(offset, 36 + dataSize, true); offset += 4;
        writeString("WAVE");
        writeString("fmt ");
        view.setUint32(offset, 16, true); offset += 4; // Subchunk1Size
        view.setUint16(offset, format, true); offset += 2; // Audio format
        view.setUint16(offset, numChannels, true); offset += 2;
        view.setUint32(offset, sampleRate, true); offset += 4;
        view.setUint32(offset, byteRate, true); offset += 4;
        view.setUint16(offset, blockAlign, true); offset += 2;
        view.setUint16(offset, bitsPerSample, true); offset += 2;
        writeString("data");
        view.setUint32(offset, dataSize, true); offset += 4;

        // Write PCM samples
        for (let i = 0; i < samples; i++) {
            for (let channel = 0; channel < numChannels; channel++) {
            let sample = audioBuffer.getChannelData(channel)[i];
            sample = Math.max(-1, Math.min(1, sample));
            view.setInt16(offset, sample * 0x7FFF, true);
            offset += 2;
            }
        }

        return buffer;
    }

    async function convertWebMBlobToWav16kHz(webmBlob) {
        // 1. Blob -> ArrayBuffer
        const arrayBuffer = await webmBlob.arrayBuffer();

        // 2. 解码为 AudioBuffer
        const audioContext = new AudioContext();
        const audioBuffer = await audioContext.decodeAudioData(arrayBuffer);

        // 3. 重采样到 16kHz
        const targetSampleRate = 16000;
        const numberOfChannels = audioBuffer.numberOfChannels;
        const durationInSeconds = audioBuffer.duration;

        const offlineContext = new OfflineAudioContext(
            numberOfChannels,
            targetSampleRate * durationInSeconds,
            targetSampleRate
        );

        const source = offlineContext.createBufferSource();
        source.buffer = audioBuffer;
        source.connect(offlineContext.destination);
        source.start(0);

        const renderedBuffer = await offlineContext.startRendering();

        // 4. 转换为 WAV 格式
        const wavBuffer = encodeWAV(renderedBuffer);

        // 5. 返回为 Blob（或可直接上传）
        return new Blob([wavBuffer], { type: 'audio/wav' });
    }

    async function objectUrlToBlob(objectUrl) {
        try {
            // 1. 发起请求获取 Blob
            const response = await fetch(objectUrl);
            
            // 2. 检查响应状态
            if (!response.ok) {
            throw new Error(`请求失败，状态码: ${response.status}`);
            }
            
            // 3. 获取 Blob 对象
            const blob = await response.blob();
            
            // 4. 验证 Blob 有效性
            if (!(blob instanceof Blob)) {
            throw new Error('获取的 Blob 无效');
            }
            
            console.log('转换成功，Blob 类型:', blob.type);
            return blob;
            
        } catch (error) {
            console.error('转换失败:', error);
            throw error; // 或返回 null/undefined
        }
    }

    function downloadWavBlob(blob, filename = 'recording.wav') {
        // 1. 创建Object URL
        const url = URL.createObjectURL(blob);
        
        // 2. 创建隐藏的<a>标签
        const a = document.createElement('a');
        a.style.display = 'none';
        a.href = url;
        a.download = filename;  // 设置下载文件名
        
        // 3. 添加到DOM并触发点击
        document.body.appendChild(a);
        a.click();
        
        // 4. 清理
        setTimeout(() => {
            document.body.removeChild(a);
            URL.revokeObjectURL(url);  // 释放内存
        }, 100);
    }


    RED.comms.subscribe("speechtotext", async function(t, objectUrl){
        const webmBlob = await objectUrlToBlob(objectUrl);

        const wavBlob = await convertWebMBlobToWav16kHz(webmBlob);
        console.log('wav: ', wavBlob);

        // downloadWavBlob(wavBlob, "testwav.wav")
        fetch("/speechtotext/wav/upload", {
            method: "POST",
            body: wavBlob,
            headers: {
                "Content-Type": "audio/wav"
            }
        }).then(res => res.text()).then(console.log);
    });
        
    RED.nodes.registerType("rdk-tools speechtotext",{
        category: "RDK Tools",
        color: "#FF804A",
        defaults: {
            name: {value:""}
        },
        inputs:1,
        outputs:1,
        align: 'right',
        icon: "stt.svg",
        paletteLabel: function() {
            return this._("rdk-speechtotext.names.speechtotext");
        },
        oneditprepare: function() {},
        label: function() {
            return this.name || this._("rdk-speechtotext.names.speechtotext");
        }
    });


})()
</script>