Update app.py

app.py

@@ -53,13 +53,13 @@ class Tango:
         for i in range(0, len(lst), n):
             yield lst[i:i + n]
         
-    def generate(self, prompt, steps=100, guidance=3, samples=3, disable_progress=True):
+    def generate(self, prompt, steps=100, guidance=3, samples=1, disable_progress=True):
         """ Genrate audio for a single prompt string. """
         with torch.no_grad():
             latents = self.model.inference([prompt], self.scheduler, steps, guidance, samples, disable_progress=disable_progress)
             mel = self.vae.decode_first_stage(latents)
             wave = self.vae.decode_to_waveform(mel)
-        return wave
+        return wave[0]
     
     def generate_for_batch(self, prompts, steps=200, guidance=3, samples=1, batch_size=8, disable_progress=True):
         """ Genrate audio for a list of prompt strings. """
@@ -83,33 +83,18 @@ tango.vae.to(device_type)
 tango.stft.to(device_type)
 tango.model.to(device_type)
     
-@spaces.GPU(duration=120)
+@spaces.GPU(duration=60)
 def gradio_generate(prompt, output_format, steps, guidance):
     output_wave = tango.generate(prompt, steps, guidance)
-    output_filename_1 = "tmp.wav"
-    wavio.write(output_filename_1, output_wave[0], rate=16000, sampwidth=2)
-    # output_wave = tango.generate_for_batch([prompt], steps, guidance, samples=3)
     # output_filename = f"{prompt.replace(' ', '_')}_{steps}_{guidance}"[:250] + ".wav"
-
-    # output_filename_1 = "tmp1.wav"
-    # wavio.write(output_filename_1, output_wave[0][0], rate=16000, sampwidth=2)
-    # output_filename_2 = "tmp2.wav"
-    # wavio.write(output_filename_2, output_wave[0][1], rate=16000, sampwidth=2)
-    # output_filename_3 = "tmp3.wav"
-    # wavio.write(output_filename_3, output_wave[0][2], rate=16000, sampwidth=2)
+    output_filename = "temp.wav"
+    wavio.write(output_filename, output_wave, rate=16000, sampwidth=2)
 
     if (output_format == "mp3"):
-        AudioSegment.from_wav("tmp1.wav").export("tmp1.mp3", format = "mp3")
-        output_filename_1 = "tmp1.mp3"
-        # AudioSegment.from_wav("tmp1.wav").export("tmp1.mp3", format = "mp3")
-        # output_filename_1 = "tmp1.mp3"
-        # AudioSegment.from_wav("tmp2.wav").export("tmp2.mp3", format = "mp3")
-        # output_filename_2 = "tmp2.mp3"
-        # AudioSegment.from_wav("tmp3.wav").export("tmp3.mp3", format = "mp3")
-        # output_filename_3 = "tmp3.mp3"
+        AudioSegment.from_wav("temp.wav").export("temp.mp3", format = "mp3")
+        output_filename = "temp.mp3"
 
-    # return [output_filename_1, output_filename_2, output_filename_3]
-    return output_filename_1
+    return output_filename
 
 # description_text = """
 # <p><a href="https://huggingface.co/spaces/declare-lab/tango/blob/main/app.py?duplicate=true"> <img style="margin-top: 0em; margin-bottom: 0em" src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a> For faster inference without waiting in queue, you may duplicate the space and upgrade to a GPU in the settings. <br/><br/>
@@ -138,11 +123,8 @@ Generate audio using Tango2 by providing a text prompt. Tango2 was built from Ta
 """
 # Gradio input and output components
 input_text = gr.Textbox(lines=2, label="Prompt")
-output_format = gr.Radio(label = "Output format", info = "The file you can download", choices = ["mp3", "wav"], value = "wav")
-output_audio_1 = gr.Audio(label="Generated Audio", type="filepath")
-# output_audio_1 = gr.Audio(label="Generated Audio #1/3", type="filepath")
-# output_audio_2 = gr.Audio(label="Generated Audio #2/3", type="filepath")
-# output_audio_3 = gr.Audio(label="Generated Audio #3/3", type="filepath")
+output_format = gr.Radio(label = "Output format", info = "The file you can dowload", choices = ["mp3", "wav"], value = "wav")
+output_audio = gr.Audio(label="Generated Audio", type="filepath")
 denoising_steps = gr.Slider(minimum=100, maximum=200, value=100, step=1, label="Steps", interactive=True)
 guidance_scale = gr.Slider(minimum=1, maximum=10, value=3, step=0.1, label="Guidance Scale", interactive=True)
 
@@ -150,8 +132,7 @@ guidance_scale = gr.Slider(minimum=1, maximum=10, value=3, step=0.1, label="Guid
 gr_interface = gr.Interface(
     fn=gradio_generate,
     inputs=[input_text, output_format, denoising_steps, guidance_scale],
-    outputs=[output_audio_1],
-    # outputs=[output_audio_1, output_audio_2, output_audio_3],
+    outputs=[output_audio],
     title="Tango 2: Aligning Diffusion-based Text-to-Audio Generations through Direct Preference Optimization",
     description=description_text,
     allow_flagging=False,