채널톡 하나1.1 TTS 인증 및 stream 엔드포인트 적용 (#6)

- 채널톡 하나1.1 TTS 인증 및 stream 엔드포인트 적용 (e8a362422024e64abe85250fda114e687bb6c6d6)

tts.py

@@ -43,6 +43,8 @@ CHANNEL_HANA_1_1_URL = os.getenv(
     "CHANNEL_HANA_1_1_URL",
     "https://ch-hana-tts.channel.io"
 )
+# 채널톡 하나1.1 TTS API Key (Bearer 토큰)
+CHANNEL_HANA_1_1_API_KEY = os.getenv("CHTTS_API_KEY")
 
 ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")
 ELEVENLABS_VOICE_ID = os.getenv("ELEVENLABS_VOICE_ID", "spzDnDXYekkOPfsxp0U3")  # Olivia (기본)
@@ -313,20 +315,33 @@ def predict_channel_tts(text: str, voice: str = "hana") -> str:
 
 
 def predict_channel_hana_1_1_tts(text: str, voice: str = "hana") -> str:
-    """채널톡 하나1.1 TTS API 호출"""
-    url = f"{CHANNEL_HANA_1_1_URL}/v1/text-to-speech/{voice}"
+    """채널톡 하나1.1 TTS API 호출
+
+    stream 엔드포인트는 raw PCM(s16le, 24kHz, mono)을 반환하므로 WAV로 감싼다.
+    """
+    if not CHANNEL_HANA_1_1_API_KEY:
+        raise ValueError("CHTTS_API_KEY 환경 변수가 설정되지 않았습니다.")
+
+    url = f"{CHANNEL_HANA_1_1_URL}/v1/text-to-speech/{voice}/stream"
 
     response = requests.post(
         url,
-        headers={"Content-Type": "application/json"},
-        json={"text": text, "output_format": "wav_24000"},
+        headers={
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {CHANNEL_HANA_1_1_API_KEY}",
+        },
+        json={"text": text, "output_format": "pcm_24000"},
         timeout=30,
     )
     response.raise_for_status()
 
-    # 임시 파일에 저장
+    # raw PCM(s16le, 24kHz, mono)을 WAV 컨테이너로 감싸 저장
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
-        f.write(response.content)
+        with wave.open(f, "wb") as wav:
+            wav.setnchannels(1)
+            wav.setsampwidth(2)  # 16-bit
+            wav.setframerate(24000)
+            wav.writeframes(response.content)
         return f.name