Text to speak generation

.gitignore

@@ -1,2 +1,4 @@
 hf/
-remote-origin.txt
+remote-origin.txt
+*.wav
+new\ audio.wav

app.py

@@ -1,7 +1,33 @@
 import gradio as gr
+from transformers import BarkModel, AutoProcessor
+import torch
+from scipy.io.wavfile import write as write_wav
+import os
 
-def greet(name):
-    return "Hello " + name + "!!"
+'''
+This app runs a text to voice transformer
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
+'''
+### Because we are using CPU we add this code: ###
+device = "cpu"
+# load in fp16
+model = BarkModel.from_pretrained("suno/bark-small").to(device)
+processor = AutoProcessor.from_pretrained("suno/bark")
+
+voice_preset = "v2/en_speaker_3"
+
+def generate_audio(text, preset, output_file_name = "bark_generation"):
+    file_name = output_file_name + ".wav"
+    inputs = processor(text, voice_preset)
+    audio_array = model.generate(**inputs)
+    audio_array = audio_array.cpu().numpy().squeeze()
+    sample_rate = model.generation_config.sample_rate
+    write_wav(file_name, sample_rate, audio_array)
+    return file_name
+
+#Presets drop down
+presets = ["v2/en_speaker_0", "v2/en_speaker_1", "v2/en_speaker_2", "v2/en_speaker_3","v2/en_speaker_4", "v2/en_speaker_5", "v2/en_speaker_6", "v2/en_speaker_9"]
+
+#Gradio interface
+iface = gr.Interface(fn=generate_audio, inputs=["text", gr.components.Dropdown(choices=presets),"text"], outputs="audio")
 iface.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio
+transformers
+torch
+accelerate
+scipy