11.25 17시

STT 컨트롤러가 텍스트 변환 결과를 프런트가 아닌 gemini에게 전달한 후 레시피 결과를 반환하도록 수정하고 stt와 gemini 도메인 통합함.

Author: hck2025

build.gradle

@@ -33,6 +33,7 @@ dependencies {
 	implementation 'io.jsonwebtoken:jjwt-api:0.11.5'
 	compileOnly 'org.projectlombok:lombok'
 	developmentOnly 'org.springframework.boot:spring-boot-devtools'
+	implementation 'com.google.cloud:google-cloud-speech:4.40.0' // google stt
 	runtimeOnly 'com.mysql:mysql-connector-j'
 	runtimeOnly 'io.jsonwebtoken:jjwt-impl:0.11.5'
 	runtimeOnly 'io.jsonwebtoken:jjwt-jackson:0.11.5'

src/main/java/com/webservice/algorithmchef/controller/SttController.java

@@ -0,0 +1,71 @@
+package com.webservice.algorithmchef.controller;
+
+import com.webservice.algorithmchef.dto.gemini.GeminiRecipeResponse;
+import com.webservice.algorithmchef.service.GeminiRecipeService;
+import com.webservice.algorithmchef.service.SpeechService;
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.http.HttpStatus;
+import org.springframework.http.ResponseEntity;
+import org.springframework.web.bind.annotation.*;
+import org.springframework.web.multipart.MultipartFile;
+
+import java.util.Collections;
+import java.util.List;
+
+@Slf4j
+@RestController
+@RequestMapping("/api")
+@CrossOrigin(origins = "http://localhost:3000")
+@RequiredArgsConstructor
+public class SttController {
+
+    private final SpeechService speechService;
+    private final GeminiRecipeService geminiRecipeService; // 레시피 서비스 객체
+
+    @PostMapping("/stt")
+    public ResponseEntity<?> stt(
+            @RequestParam("audio") MultipartFile audioFile,
+            @RequestParam("userId") String userIdStr
+    ) {
+        log.info("음성 기반 레시피 추천 요청 - UserId: {}, 파일크기: {} bytes", userIdStr, audioFile.getSize());
+
+        try {
+            // 1. 음성 -> 텍스트 변환 (STT)
+            String convertedText = speechService.stt(audioFile);
+
+            if (convertedText == null || convertedText.isEmpty()) {
+                log.warn("STT 변환 실패: 음성 인식 결과 없음");
+                return ResponseEntity.badRequest().body("음성을 인식하지 못했습니다. 다시 말씀해 주세요.");
+            }
+
+            log.info("STT 변환 결과: \"{}\"", convertedText);
+
+            // userId String -> Long 변환
+            Long userId = Long.parseLong(userIdStr);
+
+            // 음성 검색은 보통 '새로운 검색'이므로 excludedTitles는 빈 리스트로 전달
+            List<String> excludedTitles = Collections.emptyList();
+
+            // recipeservice 호출
+            List<GeminiRecipeResponse> recipes = geminiRecipeService.recommendCondition(
+                    userId,
+                    convertedText,
+                    excludedTitles
+            );
+
+            log.info("음성 추천 완료 - {}개의 레시피 반환", recipes.size());
+
+            // 3. 레시피 리스트 반환
+            return ResponseEntity.ok(recipes);
+
+        } catch (NumberFormatException e) {
+            log.error("UserId 형식 오류: {}", userIdStr, e);
+            return ResponseEntity.badRequest().body("유효하지 않은 사용자 ID입니다.");
+        } catch (Exception e) {
+            log.error("음성 레시피 추천 중 오류 발생", e);
+            return ResponseEntity.status(HttpStatus.INTERNAL_SERVER_ERROR)
+                    .body("처리 중 서버 오류가 발생했습니다: " + e.getMessage());
+        }
+    }
+}

src/main/java/com/webservice/algorithmchef/dto/gemini/GeminiRecipeResponse.java

@@ -9,8 +9,8 @@ public record GeminiRecipeResponse (
     String portions,
     String time,
     String imageUrl,
-    List<String> ingredients,
-    List<String> instructions,
+    String ingredients,
+    String instructions,
     String tip,
     String type
 ) {}

src/main/java/com/webservice/algorithmchef/service/GeminiRecipeService.java

@@ -163,8 +163,8 @@ private List<GeminiRecipeResponse> generateAndCallGemini(
                 "portions": "몇 인분"
                 "time": "예상 조리 시간",
                 "imageUrl" : "완성된 요리 사진 url 주소"
-                "ingredients": ["재료1", "재료2"],
-                "instructions": ["1. ...", "2. ..."],
+                "ingredients": "재료1", "재료2",
+                "instructions": "1. ..., 2. ...",
                 "tip": "레시피에 관한 기타 참고사항 및 추천사항",
                 "type": "gemini"
               }

src/main/java/com/webservice/algorithmchef/service/SpeechService.java

@@ -0,0 +1,58 @@
+package com.webservice.algorithmchef.service;
+
+import com.google.cloud.speech.v1.*;
+import com.google.protobuf.ByteString;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.stereotype.Service;
+import org.springframework.web.multipart.MultipartFile;
+
+import java.io.IOException;
+import java.util.List;
+
+@Slf4j
+@Service
+public class SpeechService {
+
+    private final SpeechClient speechClient;
+
+    public SpeechService() throws IOException {
+        this.speechClient = SpeechClient.create();
+    }
+
+    public String stt(MultipartFile audioFile) throws IOException {
+        byte[] data = audioFile.getBytes();
+        ByteString audioBytes = ByteString.copyFrom(data);
+
+        RecognitionConfig config = RecognitionConfig.newBuilder()
+                .setEncoding(RecognitionConfig.AudioEncoding.LINEAR16)
+                .setSampleRateHertz(16000)
+                .setLanguageCode("ko-KR")
+                .build();
+
+        RecognitionAudio audio = RecognitionAudio.newBuilder()
+                .setContent(audioBytes)
+                .build();
+
+        try {
+            RecognizeResponse response = speechClient.recognize(config, audio);
+            List<SpeechRecognitionResult> results = response.getResultsList();
+
+            if (results.isEmpty()) {
+                log.warn("STT 결과 없음 (Google API 응답이 비어있음)");
+                return "";
+            }
+
+            StringBuilder sb = new StringBuilder();
+            for (SpeechRecognitionResult result : results) {
+                SpeechRecognitionAlternative alternative = result.getAlternativesList().get(0);
+                sb.append(alternative.getTranscript());
+            }
+
+            return sb.toString();
+
+        } catch (Exception e) {
+            log.error("Google Cloud Speech API 호출 실패", e);
+            throw new IOException("Speech-to-Text fail", e);
+        }
+    }
+}