Update app.py

app.py

@@ -11,55 +11,37 @@ from whisperspeech.pipeline import Pipeline
 DEVEL=os.environ.get('DEVEL', False)
 
 title = """
-<picture>
-  <source srcset="https://huggingface.co/spaces/collabora/whisperspeech/resolve/main/dark-banner.png" media="(prefers-color-scheme: dark)" />
-  <img alt="WhisperSpeech banner with Collabora and LAION logos" src="https://huggingface.co/spaces/collabora/whisperspeech/resolve/main/light-banner.png" style="width: 60%; margin: 0 auto;" />
-</picture>
-
-# Welcome to Collabora's WhisperSpeech
-
 WhisperSpeech is an Open Source text-to-speech system built by Collabora and LAION by inverting Whisper.
-The model is fully open and you can run it on your local hardware. It's like **Stable Diffusion but for speech**
-– both powerful and easily customizable.
-
-[You can contribute to WhisperSpeech on Github.](https://github.com/collabora/WhisperSpeech)
-You can also join the discussion on Discord [![](https://dcbadge.vercel.app/api/server/FANw4rHD5E)](https://discord.gg/FANw4rHD5E)
-
-Huge thanks to [Tonic](https://huggingface.co/Tonic) who helped build this Space for WhisperSpeech.
+The model is fully open and you can run it on your local hardware.
+https://github.com/collabora/WhisperSpeech
+https://discord.gg/FANw4rHD5E
 
 ### How to Use It
-
 Write you text in the box, you can use language tags (`<en>` or `<pl>`) to create multilingual speech.
 Optionally you can upload a speech sample or give it a file URL to clone an existing voice. Check out the
 examples at the bottom of the page for inspiration.
 """
 
 footer = """
-
 ### How to use it locally
-
 ```
 pip install -U WhisperSpeech
 ```
-
 Afterwards:
-
 ```
 from whisperspeech.pipeline import Pipeline
-
 pipe = Pipeline(torch_compile=True)
 pipe.generate_to_file("output.wav", "Hello from WhisperSpeech.")
 ```
 """
 
-
 text_examples = [
     ["This is the first demo of Whisper Speech, a fully open source text-to-speech model trained by Collabora and Lion on the Juwels supercomputer.", None],
     ["World War II or the Second World War was a global conflict that lasted from 1939 to 1945. The vast majority of the world's countries, including all the great powers, fought as part of two opposing military alliances: the Allies and the Axis.", "https://upload.wikimedia.org/wikipedia/commons/7/75/Winston_Churchill_-_Be_Ye_Men_of_Valour.ogg"],
-    ["<pl>To jest pierwszy test wielojęzycznego <en>Whisper Speech <pl>, modelu zamieniającego tekst na mowę, który Collabora i Laion nauczyli na superkomputerze <en>Jewels.", None],
-    ["<en> WhisperSpeech is an Open Source library that helps you convert text to speech. <pl>Teraz także po Polsku! <en>I think I just tried saying \"now also in Polish\", don't judge me...", None],
-    # ["<de> WhisperSpeech is multi-lingual <es> y puede cambiar de idioma <hi> मध्य वाक्य में"],
-    ["<pl>To jest pierwszy test naszego modelu. Pozdrawiamy serdecznie.", None],
+    #["<pl>To jest pierwszy test wielojęzycznego <en>Whisper Speech <pl>, modelu zamieniającego tekst na mowę, który Collabora i Laion nauczyli na superkomputerze <en>Jewels.", None],
+    #["<en> WhisperSpeech is an Open Source library that helps you convert text to speech. <pl>Teraz także po Polsku! <en>I think I just tried saying \"now also in Polish\", don't judge me...", None],
+    ["<de> WhisperSpeech is multi-lingual <es> hay una vez un bar un bargochicitito <hi> मध्य वाक्य में", None],
+    #["<pl>To jest pierwszy test naszego modelu. Pozdrawiamy serdecznie.", None],
     # ["<en> The big difference between Europe <fr> et les Etats Unis <pl> jest to, że mamy tak wiele języków <uk> тут, в Європі"]
 ]
 
@@ -88,13 +70,9 @@ def generate_audio(pipe, segments, speaker, speaker_url, cps=14):
 def whisper_speech_demo(multilingual_text, speaker_audio=None, speaker_url="", cps=14):
     if len(multilingual_text) == 0:
         raise gr.Error("Please enter some text for me to speak!")
-
     segments = parse_multilingual_text(multilingual_text)
-
     audio = generate_audio(pipe, segments, speaker_audio, speaker_url, cps)
-
     return (24000, audio.T.numpy())
-
     # Did not work for me in Safari:
     # mp3 = io.BytesIO()
     # torchaudio.save(mp3, audio, 24000, format='mp3')