Merge remote-tracking branch 'origin/main'

src/main/java/com/supervision/controller/VoiceController.java

@@ -0,0 +1,33 @@
+package com.supervision.controller;
+
+import com.supervision.service.VoiceService;
+import io.swagger.annotations.Api;
+import io.swagger.annotations.ApiOperation;
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.web.bind.annotation.PostMapping;
+import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.multipart.MultipartFile;
+
+import java.io.IOException;
+
+@Slf4j
+@Api(tags = "音频处理")
+@RequestMapping("voice")
+@RequiredArgsConstructor
+public class VoiceController {
+
+    private final VoiceService voiceService;
+
+    @ApiOperation("语音转文字")
+    @PostMapping("voiceToText")
+    public String voiceToText(MultipartFile file, String sessionId) throws IOException {
+        return voiceService.voiceToText(file,sessionId);
+    }
+
+    @ApiOperation("文字转语音")
+    @PostMapping("textToVoice")
+    public String textToVoice(String text){
+        return voiceService.textToVoice(text);
+    }
+}

src/main/java/com/supervision/dto/paddlespeech/req/AsrReqDTO.java

@@ -0,0 +1,21 @@
+package com.supervision.dto.paddlespeech.req;
+
+import lombok.Data;
+
+@Data
+public class AsrReqDTO {
+
+    private final String audio;
+
+    private final String audio_format = "wav";
+
+    private final Integer sample_rate = 16000;
+
+    private final String lang = "zh_cn";
+
+    private final Integer punc = 0;
+
+    public AsrReqDTO(String audio) {
+        this.audio = audio;
+    }
+}

src/main/java/com/supervision/dto/paddlespeech/req/TtsReqDTO.java

@@ -0,0 +1,23 @@
+package com.supervision.dto.paddlespeech.req;
+
+import lombok.Data;
+
+@Data
+public class TtsReqDTO {
+
+    private final String text;
+
+    private final Integer spk_id = 0;
+
+    private final Double speed = 1.0;
+
+    private final Double volume = 1.0;
+
+    private final Integer sample_rate = 16000;
+
+    private final String save_path = "./tts.wav";
+
+    public TtsReqDTO(String text) {
+        this.text = text;
+    }
+}

src/main/java/com/supervision/dto/paddlespeech/res/AsrResultDTO.java

@@ -0,0 +1,10 @@
+package com.supervision.dto.paddlespeech.res;
+
+import lombok.Data;
+
+@Data
+public class AsrResultDTO {
+
+    private String transcription;
+
+}

src/main/java/com/supervision/dto/paddlespeech/res/PaddleSpeechResDTO.java

@@ -0,0 +1,15 @@
+package com.supervision.dto.paddlespeech.res;
+
+import lombok.Data;
+
+@Data
+public class PaddleSpeechResDTO<T> {
+
+    private Boolean success;
+
+    private Integer code;
+
+    private Object message;
+
+    private T result;
+}

src/main/java/com/supervision/dto/paddlespeech/res/TtsResultDTO.java

@@ -0,0 +1,23 @@
+package com.supervision.dto.paddlespeech.res;
+
+import lombok.Data;
+
+@Data
+public class TtsResultDTO {
+
+    private String lang;
+
+    private String spk_id;
+
+    private String speed;
+
+    private String volume;
+
+    private String sample_rate;
+
+    private String duration;
+
+    private String save_path;
+
+    private String audio;
+}

src/main/java/com/supervision/service/VoiceService.java

@@ -0,0 +1,13 @@
+package com.supervision.service;
+
+import com.supervision.vo.voice.VoiceReqVO;
+import org.springframework.web.multipart.MultipartFile;
+
+import java.io.IOException;
+
+public interface VoiceService {
+
+    String voiceToText(MultipartFile file, String sessionId) throws IOException;
+
+    String textToVoice(String text);
+}

src/main/java/com/supervision/service/impl/VoiceServiceImpl.java

@@ -0,0 +1,34 @@
+package com.supervision.service.impl;
+
+import cn.hutool.core.date.DateTime;
+import cn.hutool.core.date.DateUnit;
+import cn.hutool.core.date.DateUtil;
+import com.supervision.service.VoiceService;
+import com.supervision.util.AsrUtil;
+import com.supervision.util.TtsUtil;
+import com.supervision.vo.voice.VoiceReqVO;
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.stereotype.Service;
+import org.springframework.web.multipart.MultipartFile;
+
+import java.io.IOException;
+
+@Slf4j
+@Service
+@RequiredArgsConstructor
+public class VoiceServiceImpl implements VoiceService {
+
+    @Override
+    public String voiceToText(MultipartFile file, String sessionId) throws IOException {
+        log.info("语音转文字开始");
+        return AsrUtil.asrTransformByBytes(file.getBytes());
+    }
+
+
+    @Override
+    public String textToVoice(String text) {
+        log.info("文字转语音开始");
+        return TtsUtil.ttsTransform(text);
+    }
+}

src/main/java/com/supervision/util/AsrUtil.java

@@ -0,0 +1,46 @@
+package com.supervision.util;
+
+import cn.hutool.core.codec.Base64;
+import cn.hutool.core.util.ObjectUtil;
+import cn.hutool.http.HttpUtil;
+import cn.hutool.json.JSONUtil;
+import com.fasterxml.jackson.core.type.TypeReference;
+import com.fasterxml.jackson.databind.ObjectMapper;
+import com.supervision.dto.paddlespeech.req.AsrReqDTO;
+import com.supervision.dto.paddlespeech.res.AsrResultDTO;
+import com.supervision.dto.paddlespeech.res.PaddleSpeechResDTO;
+import com.supervision.exception.BusinessException;
+import org.springframework.core.env.Environment;
+
+public class AsrUtil {
+
+    private static final String ASR_URL = SpringBeanUtil.getBean(Environment.class).getProperty("paddle-speech.asr");
+
+    private static final ObjectMapper objectMapper = new ObjectMapper();
+
+    /**
+     * 语音转文字
+     */
+    public static String asrTransformByBytes(byte[] bytes) {
+        // 首先编码为base64编码
+        String encode = Base64.encode(bytes);
+        return asrTransformByBytes(encode);
+    }
+
+    public static String asrTransformByBytes(String voiceBase64){
+        // 这里调用Python的接口,将文字转换为语音
+        String post = HttpUtil.post(ASR_URL, JSONUtil.toJsonStr(new AsrReqDTO(voiceBase64)));
+        try {
+            PaddleSpeechResDTO<AsrResultDTO> response = objectMapper.readValue(post, new TypeReference<PaddleSpeechResDTO<AsrResultDTO>>() {
+            });
+            if (!response.getSuccess() || ObjectUtil.isEmpty(response.getResult())) {
+                throw new BusinessException("语音转换文字失败");
+            }
+            return response.getResult().getTranscription();
+
+        } catch (Exception e) {
+            throw new BusinessException("语音转换文字失败");
+        }
+
+    }
+}

src/main/java/com/supervision/util/TtsUtil.java

@@ -0,0 +1,35 @@
+package com.supervision.util;
+
+import cn.hutool.core.util.ObjectUtil;
+import cn.hutool.http.HttpUtil;
+import cn.hutool.json.JSONUtil;
+import com.fasterxml.jackson.core.type.TypeReference;
+import com.fasterxml.jackson.databind.ObjectMapper;
+import com.supervision.dto.paddlespeech.req.TtsReqDTO;
+import com.supervision.dto.paddlespeech.res.PaddleSpeechResDTO;
+import com.supervision.dto.paddlespeech.res.TtsResultDTO;
+import com.supervision.exception.BusinessException;
+import org.springframework.core.env.Environment;
+
+public class TtsUtil {
+
+    private static final String TTS_URL = SpringBeanUtil.getBean(Environment.class).getProperty("paddle-speech.tts");
+
+    private static final ObjectMapper objectMapper = new ObjectMapper();
+
+    public static String ttsTransform(String str) {
+        // 构建
+        String post = HttpUtil.post(TTS_URL, JSONUtil.toJsonStr(new TtsReqDTO(str)));
+        try {
+            PaddleSpeechResDTO<TtsResultDTO> response = objectMapper.readValue(post, new TypeReference<PaddleSpeechResDTO<TtsResultDTO>>() {
+            });
+            if (!response.getSuccess() || ObjectUtil.isEmpty(response.getResult())) {
+                throw new BusinessException("文字转换语音失败");
+            }
+            return response.getResult().getAudio();
+        } catch (Exception e) {
+            throw new BusinessException("语音转换文字失败", e);
+        }
+
+    }
+}

src/main/java/com/supervision/vo/voice/VoiceReqVO.java

@@ -0,0 +1,11 @@
+package com.supervision.vo.voice;
+
+import lombok.Data;
+
+@Data
+public class VoiceReqVO {
+
+    private String voiceBase64;
+
+    private String sessionId;
+}

src/main/resources/application.yml

@@ -44,6 +44,10 @@ spring:
 matchTool:
   url: http://192.168.10.29:8000
   scoreThreshold: 0.4
+paddle-speech:
+  # https://github.com/PaddlePaddle/PaddleSpeech/wiki/PaddleSpeech-Server-RESTful-API
+  tts: http://192.168.10.137:8090/paddlespeech/tts
+  asr: http://192.168.10.137:8090/paddlespeech/asr
 mybatis-plus:
   mapper-locations: classpath*:mapper/**/*.xml
   configuration: