Reuse generations as history prompts

README.md

@@ -6,6 +6,7 @@ https://github.com/rsxdalv/one-click-installers-tts
 This code provides a Gradio interface for generating audio from text input using the Bark TTS and Tortoise TTS models. The interface takes a text prompt as input and generates audio as output.
 
 ![Screenshot 1](./screenshots/screenshot%20(1).png)
+![Screenshot 6](./screenshots/screenshot%20(6).png)
 ![Screenshot 2](./screenshots/screenshot%20(2).png)
 ![Screenshot 3](./screenshots/screenshot%20(3).png)
 ![Screenshot 4](./screenshots/screenshot%20(4).png)

generation_tab_tortoise.py

@@ -0,0 +1,112 @@
+from gen_tortoise import generate_tortoise_n
+from models.tortoise.tortoise.utils.audio import get_voices
+
+
+import gradio as gr
+
+
+def generation_tab_tortoise():
+    with gr.Tab("Generation (Tortoise)"):
+        prompt = gr.Textbox(label="Prompt", lines=3,
+                            placeholder="Enter text here...")
+
+        with gr.Row():
+            # with gr.Box():
+            # gr.Markdown("### Voice")
+            with gr.Row():
+                voice = gr.Dropdown(
+                    choices=["random"] + list(get_voices()),
+                    value="random",
+                    # show_label=False,
+                    label="Voice"
+                )
+                # voice.style(container=False)
+                # reload_voices = gr.Button("🔁", elem_classes="btn-sm")
+                # reload_voices.style(size="sm")
+                # def reload_voices_fn():
+                #     choices =
+                #     print(choices)
+                #     return [
+                #         gr.Dropdown.update(choices=choices),
+                #     ]
+                # reload_voices.click(fn=reload_voices_fn, outputs=[voice])
+            preset = gr.Dropdown(label="Preset", choices=[
+                'ultra_fast',
+                'fast',
+                'standard',
+                'high_quality',
+            ], value="ultra_fast")
+        # Args:
+        # seed (int): The desired seed. Value must be within the inclusive range
+        #     `[-0x8000_0000_0000_0000, 0xffff_ffff_ffff_ffff]`. Otherwise, a RuntimeError
+        #     is raised. Negative inputs are remapped to positive values with the formula
+        #     `0xffff_ffff_ffff_ffff + seed`.
+        seed = gr.Textbox(label="Seed", lines=1,
+                          placeholder="Enter seed here...", value="None", visible=False)
+        cvvp_amount = gr.Slider(label="CVVP Amount",
+                                value=0.0, minimum=0.0, maximum=1.0, step=0.1)
+
+        inputs = [
+            prompt,
+            voice,
+            preset,
+            seed,
+            cvvp_amount
+        ]
+
+        with gr.Row():
+            audio_1 = gr.Audio(type="filepath", label="Generated audio")
+            audio_2 = gr.Audio(
+                type="filepath", label="Generated audio", visible=False)
+            audio_3 = gr.Audio(
+                type="filepath", label="Generated audio", visible=False)
+
+        with gr.Row():
+            image_1 = gr.Image(label="Waveform")
+            image_2 = gr.Image(label="Waveform", visible=False)
+            image_3 = gr.Image(label="Waveform", visible=False)
+
+        outputs = [audio_1, image_1]
+        outputs2 = [audio_2, image_2]
+        outputs3 = [audio_3, image_3]
+
+        with gr.Row():
+            generate3_button = gr.Button("Generate 3")
+            generate2_button = gr.Button("Generate 2")
+            generate1_button = gr.Button("Generate", variant="primary")
+
+        prompt.submit(fn=generate_tortoise_n(1),
+                      inputs=inputs, outputs=outputs)
+        generate1_button.click(fn=generate_tortoise_n(1),
+                               inputs=inputs, outputs=outputs)
+        generate2_button.click(fn=generate_tortoise_n(2), inputs=inputs,
+                               outputs=outputs + outputs2)
+        generate3_button.click(fn=generate_tortoise_n(3), inputs=inputs,
+                               outputs=outputs + outputs2 + outputs3)
+
+        def show_closure(count):
+            def show():
+                return [
+                    gr.Audio.update(visible=True),
+                    gr.Image.update(visible=True),
+                    gr.Audio.update(visible=count > 1),
+                    gr.Image.update(visible=count > 1),
+                    gr.Audio.update(visible=count > 2),
+                    gr.Image.update(visible=count > 2),
+                ]
+            return show
+
+        generate1_button.click(fn=show_closure(
+            1), outputs=outputs + outputs2 + outputs3)
+        generate2_button.click(fn=show_closure(
+            2), outputs=outputs + outputs2 + outputs3)
+        generate3_button.click(fn=show_closure(
+            3), outputs=outputs + outputs2 + outputs3)
+
+
+css_tortoise = """
+.btn-sm {
+    min-width: 3em !important;
+    flex-grow: 0 !important;
+}
+"""

history_tab.py

@@ -19,14 +19,17 @@ def open_favorites_folder():
 def save_to_favorites(json_text):
     shutil.copy(json_text["filename"], "favorites/")
     shutil.copy(json_text["filename_png"], "favorites/")
+    shutil.copy(json_text["filename_npz"], "favorites/")
     # write new json file with new filename
     json_text["filename"] = json_text["filename"].replace(
-        "outputs/", "favorites/")
+        "outputs\\", "favorites\\").replace("outputs/", "favorites/")
     json_text["filename_png"] = json_text["filename_png"].replace(
-        "outputs/", "favorites/")
+        "outputs\\", "favorites\\").replace("outputs/", "favorites/")
     json_text["filename_json"] = json_text["filename_json"].replace(
-        "outputs/", "favorites/")
-    with open(json_text["filename_json"].replace("outputs/", "favorites/"), "w") as f:
+        "outputs\\", "favorites\\").replace("outputs/", "favorites/")
+    json_text["filename_npz"] = json_text["filename_npz"].replace(
+        "outputs\\", "favorites\\").replace("outputs/", "favorites/")
+    with open(json_text["filename_json"], "w") as f:
         json.dump(json_text, f, indent=2)
     return gr.Button.update(value="Saved")
 
@@ -36,6 +39,7 @@ def delete_generation(json):
         os.remove(json["filename"])
         os.remove(json["filename_png"])
         os.remove(json["filename_json"])
+        os.remove(json["filename_npz"])
 
         return refresh()
 

save_waveform_plot.py

@@ -7,3 +7,4 @@ def save_waveform_plot(audio_array, filename_png):
     plt.plot(audio_array, color='orange')
     plt.axis("off")
     plt.savefig(filename_png)
+    plt.close()

server.py

@@ -1,8 +1,9 @@
+from generation_tab_tortoise import css_tortoise, generation_tab_tortoise
 import setup_or_recover
 import dotenv_init
 import matplotlib
 import matplotlib.pyplot as plt
-from generation_tab_bark import generation_tab_tortoise, generation_tab_bark, css_tortoise
+from generation_tab_bark import generation_tab_bark
 import gradio as gr
 import json
 from history_tab import favorites_tab, history_tab
@@ -59,9 +60,31 @@ def load_models(
     model_manager.reload_models(config)
     return gr.Button.update(value="Reload models", interactive=True)
 
-full_css = css_tortoise
-
-with gr.Blocks(css=full_css) as block:
+material_symbols_css = """
+@import url('https://fonts.googleapis.com/css2?family=Material+Symbols+Outlined:opsz,wght,FILL,GRAD@20..48,100..700,0..1,-50..200');
+
+.material-symbols-outlined {
+    font-family: 'Material Symbols Outlined' !important;
+    font-weight: normal !important;
+    font-style: normal !important;
+    font-size: 24px !important;
+    line-height: 1 !important;
+    letter-spacing: normal;
+    text-transform: none;
+    display: inline-block;
+    white-space: nowrap;
+    word-wrap: normal;
+    direction: ltr;
+    -webkit-font-feature-settings: 'liga';
+    -webkit-font-smoothing: antialiased;
+}
+"""
+
+full_css = ""
+full_css += material_symbols_css
+full_css += css_tortoise
+
+with gr.Blocks(css=full_css) as demo:
     gr.Markdown("# TTS Generation WebUI (Bark & Tortoise)")
     generation_tab_bark()
     generation_tab_tortoise()
@@ -71,10 +94,5 @@ def load_models(
 
     settings_tab(config, save_config, load_models)
 
-    block.launch(server_name='0.0.0.0')
-
-
-
-
-
-
+if __name__ == "__main__":
+    demo.launch(server_name='0.0.0.0')