一站式做网站系统蒙牛企业网站建设规划书
1、私域医疗病历问答结果展示:
最终识别结果==》张三 的 既往 疾病 是 什么
张三的既往疾病是高血压病史5年(未规律服药)。
回调函数返回的唤醒结果...:{"sst":"wakeup", "id":0, "score":2264, "bos":6030, "eos":6700 ,"keyword":"da4-fei1-da4-fei1"}
最终识别结果==》张三 的 过敏史 是 什么
张三的过敏史是青霉素过敏(皮疹反应)。
2、实现核心技术
ollama run deepseek-v2:16b
3、讯飞离线唤醒+离线合成
public static void startIvw() {Integer ret = IvwService.INSTANCE.MSPLogin(null, null, Constants.IVW_LOGIN_PARAMS); // 登录if (ret != 0) {System.out.println("唤醒登录失败...:" + ret);}String sessionId = IvwService.INSTANCE.QIVWSessionBegin(null, Constants.IVW_SSB_PARAMS, Constants.IVW_ERROR_CODE); // 开启会话if (Constants.IVW_ERROR_CODE.getValue() != 0) {System.out.println("开启唤醒会话失败...:" + Constants.IVW_ERROR_CODE.getValue());}ret = IvwService.INSTANCE.QIVWRegisterNotify(sessionId, new IvwCallback(), null); // 注册唤醒回调函数if (ret != 0) {System.out.println("注册唤醒回调函数失败...:" + ret);}try {while (true) {byte[] audioDataByteArray = new byte[Constants.IVW_FRAME_SIZE];Constants.IVW_ASR_TARGET_DATA_LINE.open(Constants.IVW_ASR_AUDIO_FORMAT);Constants.IVW_ASR_TARGET_DATA_LINE.start();int len = new AudioInputStream(Constants.IVW_ASR_TARGET_DATA_LINE).read(audioDataByteArray);if (len == -1) { // 调用麦克风时候,这段将不会被执行...Constants.IVW_AUDIO_STATUS = 4;ret = IvwService.INSTANCE.QIVWAudioWrite(sessionId, "".getBytes(), 0, Constants.IVW_AUDIO_STATUS);System.out.println("最后一帧返回的错误码:" + ret + ",即将执行退出...");break; //文件读完,跳出循环} else {// 反复调用QIVWAudioWrite写音频方法,直到音频写完为止!!!!!!!!!!!!ret = IvwService.INSTANCE.QIVWAudioWrite(sessionId, audioDataByteArray, len, Constants.IVW_AUDIO_STATUS);}Constants.IVW_AUDIO_STATUS = 2; // 中间帧if (ret != 0) {System.err.println("唤醒音频写入失败...:" + ret);}Thread.sleep(200); // 模拟人说话时间间隙,10帧的音频200ms写入一次}} catch (Exception e) {e.printStackTrace();}}
public static void startTts(String ttsText) {Constants.TTS_TEXT = ttsText;Constants.TTS_BYTE_ARRAY_OUTPUT_STREAM = new ByteArrayOutputStream(); // 初始化Constants.TTS_TOTAL_AUDIO_LENGTH = 0;Integer ret = TtsService.INSTANCE.MSPLogin(null, null, Constants.TTS_LOGIN_PARAMS); // 登录if (ret != 0) {System.out.println("合成登录失败...:" + ret);}String session_id = TtsService.INSTANCE.QTTSSessionBegin(Constants.TTS_SESSION_BEGIN_PARAMS, Constants.TTS_ERROR_CODE); // 开启合成会话if (Constants.TTS_ERROR_CODE.getValue() != 0) {System.out.println("合成开启会话失败...:" + Constants.TTS_ERROR_CODE.getValue());}ret = TtsService.INSTANCE.QTTSTextPut(session_id, Constants.TTS_TEXT, Constants.TTS_TEXT.getBytes().length, null); // 正式合成if (ret != 0) {// System.out.println("合成音频失败...:" + ret);}try { //实时播放Constants.TTS_SOURCE_DATA_LINE.open(Constants.TTS_AUDIO_FORMAT);Constants.TTS_SOURCE_DATA_LINE.start();} catch (Exception e) {e.printStackTrace();}while (true) {Pointer audioPointer = TtsService.INSTANCE.QTTSAudioGet(session_id, Constants.TTS_AUDIO_LEN, Constants.TTS_SYNTH_STATUS, Constants.TTS_ERROR_CODE); // 获取音频byte[] audioDataByteArray = null;if (audioPointer != null) {audioDataByteArray = audioPointer.getByteArray(0, Constants.TTS_AUDIO_LEN.getValue());}if (Constants.TTS_ERROR_CODE.getValue() != 0) {// System.out.println("合成获取音频失败...+:" + Constants.TTS_ERROR_CODE);// break;}if (audioDataByteArray != null) {try {Constants.TTS_SOURCE_DATA_LINE.write(audioDataByteArray, 0, Constants.TTS_AUDIO_LEN.getValue()); //实时写音频流Constants.TTS_BYTE_ARRAY_OUTPUT_STREAM.write(audioDataByteArray, 0, Constants.TTS_AUDIO_LEN.getValue()); //将数据写入字节数组的输出流,用来生成音频文件} catch (Exception e) {e.printStackTrace();}Constants.TTS_TOTAL_AUDIO_LENGTH = Constants.TTS_TOTAL_AUDIO_LENGTH + Constants.TTS_AUDIO_LEN.getValue(); //计算总音频长度,用来生成音频文件}if (Constants.TTS_SYNTH_STATUS.getValue() == 2) {// 说明音频已经取完,退出本次循环try {Constants.TTS_SOURCE_DATA_LINE.drain();Constants.TTS_SOURCE_DATA_LINE.close();Constants.TTS_BYTE_ARRAY_OUTPUT_STREAM.flush();Constants.TTS_BYTE_ARRAY_OUTPUT_STREAM.close();} catch (Exception e) {e.printStackTrace();}break;}}ret = TtsService.INSTANCE.QTTSSessionEnd(session_id, "正常退出"); //结束会话if (ret != 0) {System.out.println("合成结束会话失败...:" + ret);}ret = TtsService.INSTANCE.MSPLogout(); // 退出if (ret != 0) {System.out.println("合成退出失败...:" + ret);}}
4、JAVA通过API调用VOSK离线语音识别+DS大模型
public static void startIat() {// System.out.println("开始调用听写");AIMain.startTts("主人,您请说[p500]");while (true) {bytesRead = Constants.IVW_ASR_TARGET_DATA_LINE.read(buffer, 0, buffer.length);if (bytesRead < 0) {break;}if (recognizer.acceptWaveForm(buffer, bytesRead)) {// System.out.println(recognizer.getFinalResult());JsonParseFinal jsonParseFinal = gson.fromJson(recognizer.getFinalResult(), JsonParseFinal.class);System.err.println("最终识别结果==》" + jsonParseFinal.text);AIMain.startModel(jsonParseFinal.text);break; // 仅做一次识别} else {// System.out.println(recognizer.getPartialResult());JsonParseMid jsonParseMid = gson.fromJson(recognizer.getPartialResult(), JsonParseMid.class);System.out.println("中间识别结果==》" + jsonParseMid.partial);}}} catch (Exception e) {e.printStackTrace();}}