From 21a29a6c8a4258ba26a992a8c278a6b545b645cb Mon Sep 17 00:00:00 2001
From: dsyoon <dosangyoon@gmail.com>
Date: Fri, 30 Jan 2026 21:32:20 +0900
Subject: [PATCH] Add voice selection control

Expose a voice selector next to the save button and pass the choice to TTS so pyttsx3 can prefer a female voice.
---
 client/static/app.js        |  4 +++-
 client/static/styles.css    | 15 +++++++++++++
 client/templates/index.html |  8 ++++++-
 server/main.py              |  4 +++-
 server/tts_service.py       | 45 ++++++++++++++++++++++++++-----------
 5 files changed, 60 insertions(+), 16 deletions(-)

diff --git a/client/static/app.js b/client/static/app.js
index c3fe907..0dbe2e2 100644
--- a/client/static/app.js
+++ b/client/static/app.js
@@ -1,6 +1,7 @@
 const listEl = document.getElementById("tts-list");
 const textInput = document.getElementById("text-input");
 const saveBtn = document.getElementById("save-btn");
+const voiceSelect = document.getElementById("voice-select");
 const editBtn = document.getElementById("edit-btn");
 const deleteBtn = document.getElementById("delete-btn");
 const cancelBtn = document.getElementById("cancel-btn");
@@ -111,6 +112,7 @@ async function loadList() {
 
 async function handleSave() {
   const text = (textInput.value || "").trim();
+  const voice = (voiceSelect?.value || "male").trim();
   if (text.length < 11) {
     alert("10개 글자 이상이어야 합니다");
     return;
@@ -120,7 +122,7 @@ async function handleSave() {
   const res = await fetch("/api/tts", {
     method: "POST",
     headers: { "Content-Type": "application/json" },
-    body: JSON.stringify({ text }),
+    body: JSON.stringify({ text, voice }),
   });
 
   if (!res.ok) {
diff --git a/client/static/styles.css b/client/static/styles.css
index 9b70272..5d0e5f1 100644
--- a/client/static/styles.css
+++ b/client/static/styles.css
@@ -42,6 +42,21 @@ textarea {
   flex: 1;
 }
 
+.save-row {
+  display: flex;
+  gap: 12px;
+  align-items: center;
+}
+
+.voice-select {
+  border: 1px solid #c9c9c9;
+  border-radius: 4px;
+  padding: 10px 12px;
+  font-size: 14px;
+  background: #ffffff;
+  color: #333;
+}
+
 button {
   border: none;
   padding: 12px 18px;
diff --git a/client/templates/index.html b/client/templates/index.html
index a49ecf7..e8256e5 100644
--- a/client/templates/index.html
+++ b/client/templates/index.html
@@ -11,7 +11,13 @@
       <section class="panel left">
         <div class="panel-header">입력 텍스트</div>
         <textarea id="text-input" rows="16" placeholder="텍스트를 입력하세요"></textarea>
-        <button id="save-btn" class="primary">mp3 저장</button>
+        <div class="save-row">
+          <button id="save-btn" class="primary">mp3 변환</button>
+          <select id="voice-select" class="voice-select" aria-label="음성 선택">
+            <option value="male">남성#1</option>
+            <option value="female">여성#1</option>
+          </select>
+        </div>
         <div id="save-progress" class="progress-wrap hidden" aria-label="저장 진행률">
           <div
             id="save-progress-bar"
diff --git a/server/main.py b/server/main.py
index 55c91a0..4b93cdd 100644
--- a/server/main.py
+++ b/server/main.py
@@ -48,6 +48,7 @@ templates = Jinja2Templates(directory=str(CLIENT_DIR / "templates"))
 
 class TtsCreateRequest(BaseModel):
     text: str
+    voice: str | None = None
 
 
 class TtsDeleteRequest(BaseModel):
@@ -125,6 +126,7 @@ def api_list_tts():
 @app.post("/api/tts")
 def api_create_tts(payload: TtsCreateRequest):
     text = (payload.text or "").strip()
+    voice = (payload.voice or "").strip().lower()
     if len(text) < 11:
         raise HTTPException(status_code=400, detail="텍스트는 11글자 이상이어야 합니다.")
 
@@ -137,7 +139,7 @@ def api_create_tts(payload: TtsCreateRequest):
     mp3_path = RESOURCES_DIR / filename
 
     try:
-        text_to_mp3(text=text, mp3_path=str(mp3_path))
+        text_to_mp3(text=text, mp3_path=str(mp3_path), voice=voice)
     except Exception as exc:
         logger.exception("TTS 생성 실패")
         delete_item_by_id(tts_id)
diff --git a/server/tts_service.py b/server/tts_service.py
index b5493ec..af21598 100644
--- a/server/tts_service.py
+++ b/server/tts_service.py
@@ -111,23 +111,41 @@ def _text_to_wav_mms(text: str, wav_path: str) -> None:
     sf.write(wav_path, audio, sample_rate, subtype="PCM_16")
 
 
-def _select_korean_voice(engine: pyttsx3.Engine) -> None:
+def _select_korean_voice(engine: pyttsx3.Engine, prefer_female: bool = False) -> None:
     try:
         voices = engine.getProperty("voices") or []
     except Exception:
         return
 
-    for voice in voices:
-        lang_values = []
-        if getattr(voice, "languages", None):
-            lang_values.extend(voice.languages)
-        if getattr(voice, "id", None):
-            lang_values.append(voice.id)
-        if getattr(voice, "name", None):
-            lang_values.append(voice.name)
+    def _voice_info(v):
+        values = []
+        if getattr(v, "languages", None):
+            values.extend(v.languages)
+        if getattr(v, "id", None):
+            values.append(v.id)
+        if getattr(v, "name", None):
+            values.append(v.name)
+        return " ".join(str(x) for x in values).lower()
 
-        joined = " ".join(str(v) for v in lang_values).lower()
-        if "ko" in joined or "korean" in joined:
+    def _is_korean(info: str) -> bool:
+        return "ko" in info or "korean" in info
+
+    def _is_female(info: str) -> bool:
+        return any(token in info for token in ["female", "woman", "girl", "여성", "여자"])
+
+    if prefer_female:
+        for voice in voices:
+            info = _voice_info(voice)
+            if _is_korean(info) and _is_female(info):
+                try:
+                    engine.setProperty("voice", voice.id)
+                    return
+                except Exception:
+                    continue
+
+    for voice in voices:
+        info = _voice_info(voice)
+        if _is_korean(info):
             try:
                 engine.setProperty("voice", voice.id)
                 return
@@ -200,7 +218,7 @@ def _preprocess_text(text: str) -> str:
     return text
 
 
-def text_to_mp3(text: str, mp3_path: str) -> None:
+def text_to_mp3(text: str, mp3_path: str, voice: Optional[str] = None) -> None:
     if not text:
         raise RuntimeError("텍스트가 비어 있습니다.")
 
@@ -210,6 +228,7 @@ def text_to_mp3(text: str, mp3_path: str) -> None:
     mp3_target.parent.mkdir(parents=True, exist_ok=True)
 
     tts_engine = os.getenv("TTS_ENGINE", "pyttsx3").strip().lower()
+    voice = (voice or "").strip().lower() or None
     wav_fd, wav_path = tempfile.mkstemp(suffix=".wav")
     os.close(wav_fd)
 
@@ -226,7 +245,7 @@ def text_to_mp3(text: str, mp3_path: str) -> None:
                 engine.setProperty("volume", 1.0)
             except Exception:
                 pass
-            _select_korean_voice(engine)
+            _select_korean_voice(engine, prefer_female=voice == "female")
             # pyttsx3로 wav 생성 후 ffmpeg로 mp3 변환
             engine.save_to_file(text, wav_path)
             engine.runAndWait()