feat: 添加中文界面翻译 (i18n)

- app.py: Tab 改为中文（歌声合成、歌声转换）
- webui.py: 添加 i18n，SVS 界面全部中文化
- webui_svc.py: 使用 _i18n，SVC 界面全部中文化
- i18n_config.py: 语言配置，默认 zh

Made-with: Cursor

app.py

@@ -46,9 +46,9 @@ if __name__ == "__main__":
             '</div>'
         )
         with gr.Tabs():
-            with gr.Tab("Singing Voice Synthesis"):
+            with gr.Tab("歌声合成"):
                 render_svs_tab()
-            with gr.Tab("Singing Voice Conversion"):
+            with gr.Tab("歌声转换"):
                 render_svc_tab()
 
     page.queue()

i18n_config.py

@@ -0,0 +1,2 @@
+"""Shared i18n config for SoulX-Singer Space. Set LANG to 'zh' for Chinese, 'en' for English."""
+LANG = "zh"

webui.py

@@ -22,6 +22,45 @@ from cli.inference import build_model as build_svs_model, process as svs_process
 
 ROOT = Path(__file__).parent
 
+_I18N = dict(
+    prompt_audio_label=dict(en="Prompt audio (reference voice), max 30s", zh="Prompt 音频（参考音色），最长 30 秒"),
+    target_audio_label=dict(en="Target audio (melody / lyrics source), max 60s", zh="Target 音频（旋律/歌词来源），最长 60 秒"),
+    control_type_label=dict(en="Control type", zh="控制模式"),
+    auto_pitch_shift_label=dict(en="Auto pitch shift", zh="自动变调"),
+    generate_btn=dict(en="🎤 Generate singing voice", zh="🎤 生成歌声"),
+    advanced_accordion=dict(en="Advanced: Transcription & Metadata", zh="高级：转录与元数据"),
+    pitch_shift_label=dict(en="Pitch shift (semitones)", zh="变调（半音）"),
+    seed_label=dict(en="Seed", zh="种子"),
+    metadata_hint=dict(
+        en="Upload your own metadata files to skip automatic transcription. "
+           "You can use the [SoulX-Singer-Midi-Editor](https://huggingface.co/spaces/Soul-AILab/SoulX-Singer-Midi-Editor) to edit metadata for better alignment.",
+        zh="上传自己的 metadata 文件可跳过自动转录。可使用 [SoulX-Singer-Midi-Editor](https://huggingface.co/spaces/Soul-AILab/SoulX-Singer-Midi-Editor) 编辑 metadata 以获得更好对齐。",
+    ),
+    prompt_lyric_lang_label=dict(en="Prompt lyric language", zh="Prompt 歌词语言"),
+    target_lyric_lang_label=dict(en="Target lyric language", zh="Target 歌词语言"),
+    prompt_vocal_sep_label=dict(en="Prompt vocal separation", zh="Prompt 人声分离"),
+    target_vocal_sep_label=dict(en="Target vocal separation", zh="Target 人声分离"),
+    transcription_btn=dict(en="Run singing transcription", zh="运行歌声转录"),
+    prompt_metadata_label=dict(en="Prompt metadata", zh="Prompt 元数据"),
+    target_metadata_label=dict(en="Target metadata", zh="Target 元数据"),
+    output_audio_label=dict(en="Generated audio", zh="合成结果音频"),
+    warn_upload_both=dict(en="Please upload both prompt audio and target audio", zh="请同时上传 Prompt 与 Target 音频"),
+    warn_transcription_failed=dict(en="Transcription failed. Check your audio files.", zh="转录失败，请检查音频文件。"),
+)
+
+
+def _get_lang() -> str:
+    try:
+        from i18n_config import LANG
+        return LANG if LANG in ("zh", "en") else "zh"
+    except ImportError:
+        return "zh"
+
+
+def _i18n(key: str) -> str:
+    lang = _get_lang()
+    return _I18N.get(key, {}).get(lang, _I18N.get(key, {}).get("en", key))
+
 
 def _get_device() -> str:
     if torch.cuda.is_available():
@@ -235,7 +274,7 @@ def transcription_function(
         if isinstance(target_audio, tuple):
             target_audio = target_audio[0]
         if prompt_audio is None or target_audio is None:
-            gr.Warning(message="Please upload both prompt audio and target audio")
+            gr.Warning(message=_i18n("warn_upload_both"))
             return None, None
 
         prompt_meta_resolved = _resolve_file_path(prompt_metadata)
@@ -299,10 +338,10 @@ def synthesis_function(
             target_audio = target_audio[0]
 
         if not prompt_audio or not os.path.isfile(prompt_audio):
-            gr.Warning(message="Please upload both prompt audio and target audio")
+            gr.Warning(message=_i18n("warn_upload_both"))
             return None, gr.update(), gr.update()
         if not target_audio or not os.path.isfile(target_audio):
-            gr.Warning(message="Please upload both prompt audio and target audio")
+            gr.Warning(message=_i18n("warn_upload_both"))
             return None, gr.update(), gr.update()
 
         prompt_meta_path = _resolve_file_path(prompt_metadata)
@@ -316,7 +355,7 @@ def synthesis_function(
                 prompt_vocal_sep, target_vocal_sep,
             )
             if not p or not t:
-                gr.Warning(message="Transcription failed. Check your audio files.")
+                gr.Warning(message=_i18n("warn_transcription_failed"))
                 return None, gr.update(), gr.update()
             prompt_meta_path = p
             target_meta_path = t
@@ -369,12 +408,12 @@ def render_tab_content() -> None:
         # ── Left column: inputs & controls ──
         with gr.Column(scale=1):
             prompt_audio = gr.Audio(
-                label="Prompt audio (reference voice), max 30s",
+                label=_i18n("prompt_audio_label"),
                 type="filepath",
                 interactive=True,
             )
             target_audio = gr.Audio(
-                label="Target audio (melody / lyrics source), max 60s",
+                label=_i18n("target_audio_label"),
                 type="filepath",
                 interactive=True,
             )
@@ -383,27 +422,27 @@ def render_tab_content() -> None:
                 control_radio = gr.Radio(
                     choices=["melody", "score"],
                     value="melody",
-                    label="Control type",
+                    label=_i18n("control_type_label"),
                     scale=1,
                 )
                 auto_shift = gr.Checkbox(
-                    label="Auto pitch shift",
+                    label=_i18n("auto_pitch_shift_label"),
                     value=True,
                     interactive=True,
                     scale=1,
                 )
 
             synthesis_btn = gr.Button(
-                value="🎤 Generate singing voice",
+                value=_i18n("generate_btn"),
                 variant="primary",
                 size="lg",
             )
 
             # ── Advanced: transcription settings & metadata ──
-            with gr.Accordion("Advanced: Transcription & Metadata", open=False):
+            with gr.Accordion(_i18n("advanced_accordion"), open=False):
                 with gr.Row():
                     pitch_shift = gr.Number(
-                        label="Pitch shift (semitones)",
+                        label=_i18n("pitch_shift_label"),
                         value=0,
                         minimum=-36,
                         maximum=36,
@@ -412,21 +451,16 @@ def render_tab_content() -> None:
                         scale=1,
                     )
                     seed_input = gr.Number(
-                        label="Seed",
+                        label=_i18n("seed_label"),
                         value=12306,
                         step=1,
                         interactive=True,
                         scale=1,
                     )
-                gr.Markdown(
-                    "Upload your own metadata files to skip automatic transcription. "
-                    "You can use the [SoulX-Singer-Midi-Editor]"
-                    "(https://huggingface.co/spaces/Soul-AILab/SoulX-Singer-Midi-Editor) "
-                    "to edit metadata for better alignment."
-                )
+                gr.Markdown(_i18n("metadata_hint"))
                 with gr.Row():
                     prompt_lyric_lang = gr.Dropdown(
-                        label="Prompt lyric language",
+                        label=_i18n("prompt_lyric_lang_label"),
                         choices=[
                             ("Mandarin", "Mandarin"),
                             ("Cantonese", "Cantonese"),
@@ -437,7 +471,7 @@ def render_tab_content() -> None:
                         scale=1,
                     )
                     target_lyric_lang = gr.Dropdown(
-                        label="Target lyric language",
+                        label=_i18n("target_lyric_lang_label"),
                         choices=[
                             ("Mandarin", "Mandarin"),
                             ("Cantonese", "Cantonese"),
@@ -449,31 +483,31 @@ def render_tab_content() -> None:
                     )
                 with gr.Row():
                     prompt_vocal_sep = gr.Checkbox(
-                        label="Prompt vocal separation",
+                        label=_i18n("prompt_vocal_sep_label"),
                         value=False,
                         interactive=True,
                         scale=1,
                     )
                     target_vocal_sep = gr.Checkbox(
-                        label="Target vocal separation",
+                        label=_i18n("target_vocal_sep_label"),
                         value=True,
                         interactive=True,
                         scale=1,
                     )
                 transcription_btn = gr.Button(
-                    value="Run singing transcription",
+                    value=_i18n("transcription_btn"),
                     variant="secondary",
                     size="lg",
                 )
                 with gr.Row():
                     prompt_metadata = gr.File(
-                        label="Prompt metadata",
+                        label=_i18n("prompt_metadata_label"),
                         type="filepath",
                         file_types=[".json"],
                         interactive=True,
                     )
                     target_metadata = gr.File(
-                        label="Target metadata",
+                        label=_i18n("target_metadata_label"),
                         type="filepath",
                         file_types=[".json"],
                         interactive=True,
@@ -482,7 +516,7 @@ def render_tab_content() -> None:
         # ── Right column: output ──
         with gr.Column(scale=1):
             output_audio = gr.Audio(
-                label="Generated audio",
+                label=_i18n("output_audio_label"),
                 type="filepath",
                 interactive=False,
             )

webui_svc.py

@@ -36,8 +36,10 @@ EXAMPLE_LIST = [
 _I18N = dict(
 	display_lang_label=dict(en="Display Language", zh="显示语言"),
 	title=dict(en="## SoulX-Singer SVC", zh="## SoulX-Singer SVC"),
-	prompt_audio_label=dict(en=f"Prompt audio", zh=f"Prompt 音频"),
-	target_audio_label=dict(en=f"Target audio", zh=f"Target 音频"),
+	prompt_audio_label=dict(en="Prompt audio (reference voice)", zh="Prompt 音频（参考音色）"),
+	target_audio_label=dict(en="Target audio (to convert)", zh="Target 音频（待转换）"),
+	advanced_settings=dict(en="Advanced settings", zh="高级设置"),
+	use_fp16_label=dict(en="Use FP16", zh="使用 FP16"),
 	prompt_vocal_sep_label=dict(en="Prompt vocal separation", zh="Prompt 人声分离"),
 	target_vocal_sep_label=dict(en="Target vocal separation", zh="Target 人声分离"),
 	auto_shift_label=dict(en="Auto pitch shift", zh="自动变调"),
@@ -78,11 +80,17 @@ _I18N = dict(
     )
 )
 
-_GLOBAL_LANG: Literal["zh", "en"] = "zh"
+def _get_lang() -> str:
+	try:
+		from i18n_config import LANG
+		return LANG if LANG in ("zh", "en") else "zh"
+	except ImportError:
+		return "zh"
 
 
 def _i18n(key: str) -> str:
-	return _I18N[key][_GLOBAL_LANG]
+	lang = _get_lang()
+	return _I18N.get(key, {}).get(lang, _I18N.get(key, {}).get("en", key))
 
 
 def _print_exception(context: str) -> None:
@@ -369,39 +377,39 @@ def render_tab_content() -> None:
         # ── Left column: inputs & controls ──
         with gr.Column(scale=1):
             prompt_audio = gr.Audio(
-                label="Prompt audio (reference voice)",
+                label=_i18n("prompt_audio_label"),
                 type="filepath",
                 interactive=True,
             )
             target_audio = gr.Audio(
-                label="Target audio (to convert)",
+                label=_i18n("target_audio_label"),
                 type="filepath",
                 interactive=True,
             )
 
             run_btn = gr.Button(
-                value="🎤 Singing Voice Conversion",
+                value=_i18n("run_btn"),
                 variant="primary",
                 size="lg",
             )
 
-            with gr.Accordion("Advanced settings", open=False):
+            with gr.Accordion(_i18n("advanced_settings"), open=False):
                 with gr.Row():
-                    prompt_vocal_sep = gr.Checkbox(label="Prompt vocal separation", value=False, scale=1)
-                    target_vocal_sep = gr.Checkbox(label="Target vocal separation", value=True, scale=1)
+                    prompt_vocal_sep = gr.Checkbox(label=_i18n("prompt_vocal_sep_label"), value=False, scale=1)
+                    target_vocal_sep = gr.Checkbox(label=_i18n("target_vocal_sep_label"), value=True, scale=1)
                 with gr.Row():
-                    auto_shift = gr.Checkbox(label="Auto pitch shift", value=True, scale=1)
-                    auto_mix_acc = gr.Checkbox(label="Auto mix accompaniment", value=True, scale=1)
+                    auto_shift = gr.Checkbox(label=_i18n("auto_shift_label"), value=True, scale=1)
+                    auto_mix_acc = gr.Checkbox(label=_i18n("auto_mix_acc_label"), value=True, scale=1)
                 with gr.Row():
-                    use_fp16 = gr.Checkbox(label="Use FP16", value=True, scale=1)
-                pitch_shift = gr.Slider(label="Pitch shift (semitones)", value=0, minimum=-36, maximum=36, step=1)
-                n_step = gr.Slider(label="diffusion steps", value=32, minimum=1, maximum=200, step=1)
-                cfg = gr.Slider(label="cfg scale", value=1.0, minimum=0.0, maximum=10.0, step=0.1)
-                seed_input = gr.Slider(label="Seed", value=42, minimum=0, maximum=10000, step=1)
+                    use_fp16 = gr.Checkbox(label=_i18n("use_fp16_label"), value=True, scale=1)
+                pitch_shift = gr.Slider(label=_i18n("pitch_shift_label"), value=0, minimum=-36, maximum=36, step=1)
+                n_step = gr.Slider(label=_i18n("n_step_label"), value=32, minimum=1, maximum=200, step=1)
+                cfg = gr.Slider(label=_i18n("cfg_label"), value=1.0, minimum=0.0, maximum=10.0, step=0.1)
+                seed_input = gr.Slider(label=_i18n("seed_label"), value=42, minimum=0, maximum=10000, step=1)
 
         # ── Right column: output ──
         with gr.Column(scale=1):
-            output_audio = gr.Audio(label="Generated audio", type="filepath", interactive=False)
+            output_audio = gr.Audio(label=_i18n("output_audio_label"), type="filepath", interactive=False)
             svc_state = gr.State(value=None)
             gr.Examples(
                 examples=EXAMPLE_LIST,