Spaces:
Runtime error
Runtime error
Commit
·
bc01a52
1
Parent(s):
c48b863
Update app.py
Browse files
app.py
CHANGED
@@ -208,79 +208,27 @@ def predict_song_from_yt(
|
|
208 |
|
209 |
description = f"""
|
210 |
|
211 |
-
## <center>🏞️ -
|
212 |
|
213 |
-
### <center>🌟 -
|
214 |
|
215 |
-
### <center
|
216 |
|
217 |
-
<center>💡 - 如何使用此程序:在页面上方选择“从B站视频上传”模块,填写视频网址和视频起止时间后,点击“submit”按键即可!您还可以点击页面最下方的示例快速预览效果</center>
|
218 |
-
|
219 |
-
<h1 align="center"><a href="http://www.talktalkai.com"><img src="https://y.qq.com/music/photo_new/T001R300x300M0000025Gr0r2OXvrn_2.jpg", alt="talktalkai" border="0" style="margin: 0 auto; height: 200px;" /></a> </h1>
|
220 |
-
|
221 |
-
|
222 |
-
""".strip()
|
223 |
-
|
224 |
-
article = """
|
225 |
-
<p style='text-align: center'> 注意❗:请不要生成会对个人以及组织造成侵害的内容,此程序仅供科研、学习及个人娱乐使用。
|
226 |
-
</p>
|
227 |
-
<p style='text-align: center'> 🌊🎶🏞️ - 江水东流急,滔滔无尽声。 明·顾璘
|
228 |
-
</p>
|
229 |
""".strip()
|
230 |
|
231 |
|
232 |
-
interface_mic = gr.Interface(
|
233 |
-
predict,
|
234 |
-
inputs=[
|
235 |
-
gr.Dropdown(speakers, value=speakers[0], label="🎤AI歌手🎶 - 🌟一清清清🌟"),
|
236 |
-
gr.Audio(type="filepath", source="microphone", label="请用麦克风上传您想转换的歌曲"),
|
237 |
-
gr.Slider(-12, 12, value=0, step=1, label="变调 (默认为0;有正负值,+2为升高两个key)"),
|
238 |
-
gr.Checkbox(False, label="是否开启自动f0预测", info="勾选即为开启;配合聚类模型f0预测效果更好,仅限语音转换时使用", visible=False),
|
239 |
-
gr.Slider(0.0, 1.0, value=default_cluster_infer_ratio, step=0.1, label="聚类模型混合比例", info="0-1之间,0即不启用聚类。使用聚类模型能提升音色相似度,但会导致咬字下降 (如果使用,建议0.5左右)"),
|
240 |
-
gr.Slider(0.0, 1.0, value=0.4, step=0.1, label="noise scale (建议保持不变)", visible=False),
|
241 |
-
gr.Dropdown(
|
242 |
-
choices=["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"],
|
243 |
-
value=default_f0_method,
|
244 |
-
label="模型推理方法 (crepe推理效果最好)", visible=False
|
245 |
-
),
|
246 |
-
],
|
247 |
-
outputs="audio",
|
248 |
-
title="🌊💕🎶 - 滔滔AI+音乐:可从B站直接上传素材,无需分离背景音",
|
249 |
-
description=description,
|
250 |
-
article=article,
|
251 |
-
)
|
252 |
-
interface_file = gr.Interface(
|
253 |
-
predict,
|
254 |
-
inputs=[
|
255 |
-
gr.Dropdown(speakers, value=speakers[0], label="🎤AI歌手🎶 - 🌟一清清清🌟"),
|
256 |
-
gr.Audio(type="filepath", source="upload", label="请上传您想转换的歌曲 (仅人声部分)"),
|
257 |
-
gr.Slider(-12, 12, value=0, step=1, label="变调 (默认为0;有正负值,+2为升高两个key)"),
|
258 |
-
gr.Checkbox(False, label="是否开启自动f0预测", info="勾选即为开启;配合聚类模型f0预测效果更好,仅限语音转换时使用", visible=False),
|
259 |
-
gr.Slider(0.0, 1.0, value=default_cluster_infer_ratio, step=0.1, label="聚类模型混合比例", info="0-1之间,0即不启用聚类。使用聚类模型能提升音色相似度,但会导致咬字下降 (如果使用,建议0.5左右)"),
|
260 |
-
gr.Slider(0.0, 1.0, value=0.4, step=0.1, label="noise scale (建议保持不变)", visible=False),
|
261 |
-
gr.Dropdown(
|
262 |
-
choices=["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"],
|
263 |
-
value=default_f0_method,
|
264 |
-
label="模型推理方法 (crepe推理效果最好)", visible=False
|
265 |
-
),
|
266 |
-
],
|
267 |
-
outputs="audio",
|
268 |
-
title="🌊💕🎶 - 滔滔AI+音乐:可从B站直接上传素材,无需分离背景音",
|
269 |
-
description=description,
|
270 |
-
article=article,
|
271 |
-
)
|
272 |
interface_yt = gr.Interface(
|
273 |
predict_song_from_yt,
|
274 |
inputs=[
|
275 |
gr.Textbox(
|
276 |
-
label="
|
277 |
),
|
278 |
-
gr.Number(value=0, label="
|
279 |
-
gr.Number(value=15, label="
|
280 |
-
gr.Dropdown(speakers, value=speakers[0], label="🎤AI
|
281 |
-
gr.Slider(-12, 12, value=0, step=1, label="
|
282 |
gr.Checkbox(False, label="是否开启自动f0预测", info="勾选即为开启;配合聚类模型f0预测效果更好,仅限语音转换时使用", visible=False),
|
283 |
-
gr.Slider(0.0, 1.0, value=default_cluster_infer_ratio, step=0.1, label="聚类模型混合比例", info="0-1之间,0即不启用聚类。使用聚类模型能提升音色相似度,但会导致咬字下降"),
|
284 |
gr.Slider(0.0, 1.0, value=0.4, step=0.1, label="noise scale (建议保持不变)", visible=False),
|
285 |
gr.Dropdown(
|
286 |
choices=["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"],
|
@@ -288,17 +236,17 @@ interface_yt = gr.Interface(
|
|
288 |
label="模型推理方法 (crepe推理效果最好)", visible=False
|
289 |
),
|
290 |
],
|
291 |
-
outputs=[gr.Audio(label="
|
292 |
-
title="🌊💕🎶 -
|
293 |
description=description,
|
294 |
article=article,
|
295 |
examples=[
|
296 |
-
["https://www.
|
297 |
],
|
298 |
)
|
299 |
interface = gr.TabbedInterface(
|
300 |
-
[interface_yt
|
301 |
-
["📺 -
|
302 |
)
|
303 |
|
304 |
|
|
|
208 |
|
209 |
description = f"""
|
210 |
|
211 |
+
## <center>🏞️ - TalkTalkAI - Generative AI Text to Speech & Singing Voice Cloning</center>
|
212 |
|
213 |
+
### <center>🌟 - The singer who collaborates with TalkTalkAI[一清清清](https://space.bilibili.com/22960772?spm_id_from=333.337.0.0)</center>
|
214 |
|
215 |
+
### <center>🤗 - Stay tuned. The best is yet to come. Contact us: talktalkai.kevin@gmail.com</center>
|
216 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
217 |
""".strip()
|
218 |
|
219 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
220 |
interface_yt = gr.Interface(
|
221 |
predict_song_from_yt,
|
222 |
inputs=[
|
223 |
gr.Textbox(
|
224 |
+
label="YouTube URL or ID", info="A YouTube URL (or ID) to a song on YouTube you want to clone from"
|
225 |
),
|
226 |
+
gr.Number(value=0, label="Start Time (seconds)"),
|
227 |
+
gr.Number(value=15, label="End Time (seconds)"),
|
228 |
+
gr.Dropdown(speakers, value=speakers[0], label="🎤 AI Singer"),
|
229 |
+
gr.Slider(-12, 12, value=0, step=1, label="Transpose (Semitones)"),
|
230 |
gr.Checkbox(False, label="是否开启自动f0预测", info="勾选即为开启;配合聚类模型f0预测效果更好,仅限语音转换时使用", visible=False),
|
231 |
+
gr.Slider(0.0, 1.0, value=default_cluster_infer_ratio, step=0.1, label="聚类模型混合比例", info="0-1之间,0即不启用聚类。使用聚类模型能提升音色相似度,但会导致咬字下降", visible=False),
|
232 |
gr.Slider(0.0, 1.0, value=0.4, step=0.1, label="noise scale (建议保持不变)", visible=False),
|
233 |
gr.Dropdown(
|
234 |
choices=["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"],
|
|
|
236 |
label="模型推理方法 (crepe推理效果最好)", visible=False
|
237 |
),
|
238 |
],
|
239 |
+
outputs=[gr.Audio(label="With BGM🎵"), gr.Audio(label="Without BGM🎤")],
|
240 |
+
title="🌊💕🎶 - TalkTalkAI",
|
241 |
description=description,
|
242 |
article=article,
|
243 |
examples=[
|
244 |
+
["https://www.youtube.com/watch?v=cn4M-fH08XY", 0, 11, speakers[0], 0, False, default_cluster_infer_ratio, 0.4, default_f0_method],
|
245 |
],
|
246 |
)
|
247 |
interface = gr.TabbedInterface(
|
248 |
+
[interface_yt],
|
249 |
+
["📺 - Clone Song From YouTube],
|
250 |
)
|
251 |
|
252 |
|