Modularize Audiocraft

README.md

@@ -30,29 +30,32 @@
 https://rsxdalv.github.io/bark-speaker-directory/
 
 ## Changelog
+July 5:
+* Improved v5 installer - faster and more reliable
+
 July 2:
-* Upgrade bark settings #59
+* Upgrade bark settings https://github.com/rsxdalv/tts-generation-webui/pull/59
 
 July 1:
-* Studio-tab #58
+* Studio-tab https://github.com/rsxdalv/tts-generation-webui/pull/58
 
 Jun 29:
-* Tortoise new params #54
+* Tortoise new params https://github.com/rsxdalv/tts-generation-webui/pull/54
 
 Jun 27:
-* Fix eager loading errors, refactor #50
+* Fix eager loading errors, refactor https://github.com/rsxdalv/tts-generation-webui/pull/50
 
 Jun 20
-* Tortoise: proper long form generation files #46
+* Tortoise: proper long form generation files https://github.com/rsxdalv/tts-generation-webui/pull/46
 
 Jun 19
-* Tortoise-upgrade #45
+* Tortoise-upgrade https://github.com/rsxdalv/tts-generation-webui/pull/45
 
 June 18:
 * Update to newest audiocraft, add longer generations
 
 Jun 14:
-* add vocos wav tab #42
+* add vocos wav tab https://github.com/rsxdalv/tts-generation-webui/pull/42
 
 June 5:
 * Fix "Save to Favorites" button on bark generation page, clean up console (v4.1.1)

requirements.txt

@@ -4,7 +4,5 @@ python-dotenv==1.0.0
 soundfile==0.12.1; sys_platform == 'win32' # torchaudio
 # sox # torchaudio for linux
 suno-bark @ git+https://github.com/suno-ai/bark@7280e4e#egg=suno-bark # MIT License
-# audiocraft==0.0.1 # MIT License
-audiocraft @ git+https://git@github.com/facebookresearch/audiocraft@d874966#egg=audiocraft # MIT License
 vocos==0.0.2 # MIT License
 tortoise @ git+https://github.com/rsxdalv/tortoise-tts@f04f95e30245c85996be3af068cba4890952b1a1#egg=tortoise # Apache 2.0

requirements_audiocraft.txt

@@ -0,0 +1,2 @@
+# audiocraft==0.0.1 # MIT License
+audiocraft @ git+https://git@github.com/facebookresearch/audiocraft@d874966#egg=audiocraft # MIT License

server.py

@@ -3,22 +3,12 @@
 import src.utils.dotenv_init as dotenv_init
 import gradio as gr
 
-from src.bark.clone.tab_voice_clone_error import tab_voice_clone_error
+from src.config.load_config import default_config
+from src.config.config import config
+
 from src.css.css import full_css
 from src.Joutai import Joutai
-from src.musicgen.musicgen_tab import generation_tab_musicgen
 from src.history_tab.collections_directories_atom import collections_directories_atom
-from src.config.save_config_gradio import save_config_gradio
-from src.tortoise.generation_tab_tortoise import generation_tab_tortoise
-from src.config.load_config import default_config
-from src.settings_tab_gradio import settings_tab_gradio
-from src.bark.generation_tab_bark import generation_tab_bark
-from src.history_tab.main import history_tab
-from src.bark.settings_tab_bark import settings_tab_bark
-from src.config.config import config
-from src.history_tab.voices_tab import voices_tab
-from src.vocos.vocos_tabs import vocos_tabs
-from src.studio.studio_tab import simple_remixer_tab
 
 setup_or_recover.dummy()
 dotenv_init.init()
@@ -47,18 +37,39 @@ def reload_config_and_restart_ui():
 ) as demo:
     gr.Markdown("# TTS Generation WebUI (Bark, MusicGen, Tortoise)")
     with Joutai.singleton.tabs:
+        from src.tortoise.generation_tab_tortoise import generation_tab_tortoise
+        from src.settings_tab_gradio import settings_tab_gradio
+        from src.bark.generation_tab_bark import generation_tab_bark
+        from src.history_tab.main import history_tab
+        from src.bark.settings_tab_bark import settings_tab_bark
+        from src.history_tab.voices_tab import voices_tab
+        from src.vocos.vocos_tabs import vocos_tabs
+        from src.studio.studio_tab import simple_remixer_tab
+
         register_use_as_history_button = generation_tab_bark()
 
         try:
             from src.bark.clone.tab_voice_clone import tab_voice_clone
 
             tab_voice_clone(register_use_as_history_button)
         except Exception as e:
+            from src.bark.clone.tab_voice_clone_error import tab_voice_clone_error
+
             tab_voice_clone_error(e)
             print("Failed to load voice clone demo")
             print(e)
 
-        generation_tab_musicgen()
+        try:
+            from src.musicgen.musicgen_tab import generation_tab_musicgen
+
+            generation_tab_musicgen()
+        except Exception as e:
+            from src.musicgen.musicgen_tab_error import musicgen_tab_error
+
+            musicgen_tab_error(e)
+            print("Failed to load musicgen demo")
+            print(e)
+
         vocos_tabs()
         generation_tab_tortoise()
 
@@ -71,9 +82,7 @@ def reload_config_and_restart_ui():
         voices_tab(register_use_as_history_button)
 
         settings_tab_bark()
-        settings_tab_gradio(
-            save_config_gradio, reload_config_and_restart_ui, gradio_interface_options
-        )
+        settings_tab_gradio(reload_config_and_restart_ui, gradio_interface_options)
         remixer_input = simple_remixer_tab()
     Joutai.singleton.tabs.render()
 
@@ -100,7 +109,12 @@ def print_pretty_options(options):
     print("Gradio server authentication enabled")
 print_pretty_options(gradio_interface_options)
 
-if __name__ == "__main__":
+
+def start_server():
     demo.queue(
         concurrency_count=gradio_interface_options.get("concurrency_count", 5),
     ).launch(**gradio_interface_options)
+
+
+if __name__ == "__main__":
+    start_server()

src/bark/clone/tab_voice_clone_error.py

@@ -25,11 +25,9 @@ def install_bark_voice_clone():
     if process.returncode == 0:
         print("Successfully installed bark voice clone")
         yield "Successfully installed bark voice clone, please restart the webui"
-        return
     else:
         print("Failed to install bark voice clone")
         yield "Failed to install bark voice clone"
-        return
 
 
 def tab_voice_clone_error(e: Exception):
@@ -38,7 +36,7 @@ def tab_voice_clone_error(e: Exception):
         gr.Markdown(f"Error: {e}")
         gr.Markdown("Please install the requirements_bark_hubert_quantizer.txt file")
         gr.Markdown("Please check the console for more information")
-        install_btn = gr.Button("install bark voice clone")
+        install_btn = gr.Button("Install Bark Voice Clone")
         gr.Markdown("Installation console:")
         console_text = gr.HTML()
         install_btn.click(

src/config/load_config.py

@@ -8,9 +8,9 @@
         "coarse_use_gpu": True,
         "coarse_use_small": True,
         "fine_use_gpu": True,
-        "fine_use_small": False,
+        "fine_use_small": True,
         "codec_use_gpu": True,
-        "load_models_on_startup": True,
+        "load_models_on_startup": False,
     },
     "gradio_interface_options": {
         "inline": False,
@@ -38,7 +38,7 @@
         "file_directories": None,
         "_frontend": True,
     },
-    "load_models_on_startup": True,
+    "load_models_on_startup": False,
 }
 
 

src/musicgen/musicgen_tab.py

@@ -197,26 +197,25 @@ def generate(params: MusicGenGeneration, melody_in: Optional[Tuple[int, np.ndarr
     ]
 
 
-musicgen_atom = gr.JSON(
-    # visible=True,
-    visible=False,
-    value={
-        "text": "",
-        "melody": None,
-        "model": "melody",
-        "duration": 10,
-        "topk": 250,
-        "topp": 0,
-        "temperature": 1.0,
-        "cfg_coef": 3.0,
-        "seed": -1,
-    },
-)
-
 
 def generation_tab_musicgen():
     with gr.Tab("MusicGen") as tab:
-        musicgen_atom.render()
+        musicgen_atom = gr.JSON(
+            # visible=True,
+            visible=False,
+            value={
+                "text": "",
+                "melody": None,
+                "model": "melody",
+                "duration": 10,
+                "topk": 250,
+                "topp": 0,
+                "temperature": 1.0,
+                "cfg_coef": 3.0,
+                "seed": -1,
+            },
+        )
+        # musicgen_atom.render()
         gr.Markdown(f"""Audiocraft version: {AUDIOCRAFT_VERSION}""")
         with gr.Row():
             with gr.Column():

src/musicgen/musicgen_tab_error.py

@@ -0,0 +1,50 @@
+import gradio as gr
+
+
+def install_musicgen():
+    import subprocess
+
+    process = subprocess.Popen(
+        "pip install -r requirements_audiocraft.txt",
+        shell=True,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.STDOUT,
+        universal_newlines=True,
+    )
+
+    # Stream the output to the console
+    for line in process.stdout:  # type: ignore
+        print(line, end="")
+        yield line
+
+    # Wait for the process to finish
+    process.wait()
+
+    # Check if the process was successful
+    if process.returncode == 0:
+        print("Successfully installed MusicGen")
+        yield "Successfully installed MusicGen, please restart the webui"
+    else:
+        print("Failed to install MusicGen")
+        yield "Failed to install MusicGen"
+
+
+def musicgen_tab_error(e: Exception):
+    with gr.Tab("MusicGen (!)"):
+        gr.Markdown("Failed to load voice clone demo")
+        gr.Markdown(f"Error: {e}")
+        gr.Markdown("Please install the requirements_bark_hubert_quantizer.txt file")
+        gr.Markdown("Please check the console for more information")
+        install_btn = gr.Button("Install MusicGen")
+        gr.Markdown("Installation console:")
+        console_text = gr.HTML()
+        install_btn.click(
+            install_musicgen,
+            outputs=[console_text],
+        )
+
+
+if __name__ == "__main__":
+    with gr.Blocks() as demo:
+        musicgen_tab_error(Exception("Test"))
+    demo.launch()

src/settings_tab_gradio.py

@@ -1,9 +1,9 @@
-from typing import Any, Callable, Dict, List
+from typing import Any, Callable, Dict
 import gradio as gr
+from src.config.save_config_gradio import save_config_gradio
 
 
 def settings_tab_gradio(
-    save_config: Callable[[List[str], List[Any]], str],
     reload_config_and_restart_ui: Callable[[], None],
     gradio_interface_options: Dict[str, Any],
 ):
@@ -155,7 +155,7 @@ def settings_tab_gradio(
                 # Map over the UI elements
                 for i in inputs:
                     i.change(
-                        fn=lambda *input_values: save_config(keys, input_values),
+                        fn=lambda *input_values: save_config_gradio(keys, input_values),
                         inputs=inputs,
                         outputs=[save_beacon],
                     )