Update app.py

app.py

@@ -149,21 +149,19 @@ def model_inference(
 
 
 html_intro = """
-<div>
-    <h2 style="font-size: 22px;margin-left: 0px;">语音理解模型： SenseVoice-Small</h2>
-    <p style="font-size: 18px;margin-left: 20px;">
-SenseVoice-Small 是一种仅包含编码器的语音基础模型，专为快速语音理解而设计。它包含多种功能，包括自动语音识别 (ASR)、口语识别 (LID)、语音情感识别 (SER) 和声学事件检测 (AED)。 SenseVoice-Small支持中文、英语、粤语、日语、韩语等多语言识别。此外，它还提供极低的推理延迟，执行速度比 Whisper-small 快 7 倍，比 Whisper-large 快 17 倍。</p>
-    <h2 style="font-size: 22px;margin-left: 0px;">用法</h2> <p style="font-size: 18px;margin-left: 20px;">上传音频文件或通过麦克风输入，然后选择任务和语言。音频被转录成相应的文本以及相关的情绪（😊快乐，😡愤怒/兴奋，😔悲伤）和声音事件类型（😀笑声，🎼音乐，👏掌声，🤧咳嗽和打喷嚏，😔哭泣）。事件标签放在文本的前面，情感放在文本的后面。</p>
-	<h2 style="font-size: 22px;margin-left: 0px;">目录</h2>
-	<p style="font-size: 18px;margin-left: 20px;"><a href="https://github.com/FunAudioLLM/SenseVoice" target="_blank">SenseVoice</a>：多语言语音理解模型</p>
-	<p style="font-size: 18px;margin-left: 20px;"><a href="https://github.com/modelscope/FunASR" target="_blank">FunASR</a>：基本语音识别工具包</p>
-	<p style="font-size: 18px;margin-left: 20px;"><a href="https://github.com/FunAudioLLM/CosyVoice" target="_blank">CosyVoice</a>：高质量多语言TTS模型</p>
-</div>
-<div style="text-align: center; font-size: 15px; font-weight: bold; color: red; margin-bottom: 20px;">
-        ⚠️ 该演示仅供学术研究和体验使用。
+<div style="text-align: center; font-family: var(--font-sans);">
+    <h1 style="font-size: 28px;">SenseVoice-Small 语音识别</h1>
+    <p>SenseVoice 是具有音频理解能力的音频基础模型，包括语音识别（ASR）、语种识别（LID）、语音情感识别（SER）和声学事件分类（AEC）或声学事件检测（AED）。</p>
+    <p style="font-size: small; color: #888;">支持 MP3, WAV, FLAC, M4A 等常见音频格式。</p>
 </div>
 """
 
+custom_css = """
+body, .gradio-container {
+    font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, 'Open Sans', 'Helvetica Neue', sans-serif;
+}
+"""
+
 lang_options = [
     ("自动检测", "auto"),
     ("中文", "zh"),
@@ -173,55 +171,30 @@ lang_options = [
     ("한국어", "ko"),
     ("无语音", "nospeech"),
 ]
-example_list = [
-    [f"{model.model_path}/example/zh.mp3", "auto"],
-    [f"{model.model_path}/example/en.mp3", "auto"],
-    [f"{model.model_path}/example/yue.mp3", "auto"],
-    [f"{model.model_path}/example/ja.mp3", "auto"],
-    [f"{model.model_path}/example/ko.mp3", "auto"],
-    [f"{model.model_path}/example/rich_1.wav", "auto"],
-    [f"{model.model_path}/example/rich_2.wav", "auto"],
-    [f"{model.model_path}/example/longwav_1.wav", "auto"],
-]
 
-with gr.Blocks(theme=gr.themes.Soft()) as demo:
+with gr.Blocks(theme=gr.themes.Soft(), css=custom_css) as demo:
     gr.HTML(html_intro)
     with gr.Row():
         with gr.Column(scale=2):
-            audio_inputs = gr.Audio(label="上传音频或使用麦克风")
+            audio_inputs = gr.Audio(label="上传/录制音频")
+            language_inputs = gr.Dropdown(
+                choices=lang_options, value="auto", label="源语言"
+            )
+            output_format_dropdown = gr.Dropdown(
+                choices=["纯净文本", "原始富文本", "Emoji 格式"],
+                value="纯净文本",
+                label="输出格式",
+            )
             with gr.Accordion("高级设置", open=False):
-                output_format_dropdown = gr.Dropdown(
-                    choices=["Emoji格式化", "纯净文本", "原始富文本"],
-                    value="Emoji格式化",
-                    label="选择输出格式",
-                )
-                language_inputs = gr.Dropdown(
-                    choices=lang_options, value="auto", label="识别语言"
-                )
-                use_itn_checkbox = gr.Checkbox(
-                    value=True, label="启用标点和逆文本正规化"
-                )
-                merge_vad_checkbox = gr.Checkbox(
-                    value=True, label="合并VAD切分的音频片段"
-                )
+                use_itn_checkbox = gr.Checkbox(value=True, label="自动添加标点与格式化")
+                merge_vad_checkbox = gr.Checkbox(value=True, label="优化长音频断句")
                 merge_length_slider = gr.Slider(
-                    minimum=5,
-                    maximum=30,
-                    value=15,
-                    step=1,
-                    label="合并后片段最大长度(秒)",
-                )
-                ban_emo_unk_checkbox = gr.Checkbox(
-                    value=False, label="禁用未知情感标签"
+                    minimum=5, maximum=30, value=15, step=1, label="断句最大长度（秒）"
                 )
+                ban_emo_unk_checkbox = gr.Checkbox(value=False, label="强制情感分类")
             fn_button = gr.Button("开始识别", variant="primary")
         with gr.Column(scale=3):
             text_outputs = gr.Textbox(label="识别结果", lines=20)
-            gr.Examples(
-                examples=example_list,
-                inputs=[audio_inputs, language_inputs],
-                label="示例音频 (点击即可试用)",
-            )
     fn_button.click(
         fn=model_inference,
         inputs=[