app.py

#!/usr/bin/env python
# coding: utf-8

# In[1]:


import gradio as gr

import torch


# In[11]:


import torch.nn as nn


# In[ ]:





# In[ ]:





# In[12]:


class Conv1DNet(nn.Module):
    def __init__(self):
        super(Conv1DNet, self).__init__()
        
        # First Conv1D layer
        self.conv1 = nn.Conv1d(in_channels=1, out_channels=8, kernel_size=13, stride=1, padding=1)
        self.relu1 = nn.ReLU()
        self.pool1 = nn.MaxPool1d(kernel_size=3)
        self.dropout1 = nn.Dropout(p=0.3)
        
        # Second Conv1D layer
        self.conv2 = nn.Conv1d(in_channels=8, out_channels=16, kernel_size=11, stride=1, padding=1)
        self.relu2 = nn.ReLU()
        self.pool2 = nn.MaxPool1d(kernel_size=3)
        self.dropout2 = nn.Dropout(p=0.3)
        
        # Third Conv1D layer
        self.conv3 = nn.Conv1d(in_channels=16, out_channels=32, kernel_size=9, stride=1, padding=1)
        self.relu3 = nn.ReLU()
        self.pool3 = nn.MaxPool1d(kernel_size=3)
        self.dropout3 = nn.Dropout(p=0.3)
        # forth Conv1D layer
        self.conv4 = nn.Conv1d(in_channels=32, out_channels=64, kernel_size=7, stride=1, padding=1)
        self.relu4 = nn.ReLU()
        self.pool4 = nn.MaxPool1d(kernel_size=3)
        self.dropout4 = nn.Dropout(p=0.3)
        
        
        # Fully connected layer
        self.fc1 = nn.Linear(in_features=6144, out_features=10)
        #self.fc2 = nn.Linear(in_features=256, out_features=10)

    def forward(self, x):
        # Pass the input through each layer
        batch_size=x.size(0)
        #print(x.size())
        #print(x)
        x = self.conv1(x)
        x = self.relu1(x)
        #print(x.size())
        x = self.pool1(x)
        x = self.dropout1(x)
        x = self.conv2(x)
        x = self.relu2(x)
        x = self.pool2(x)
        x = self.dropout2(x)
        x = self.conv3(x)
        x = self.relu3(x)
        x = self.pool3(x)
        x = self.dropout3(x)
        x = self.conv4(x)
        x = self.relu4(x)
        x = self.pool4(x)
        x = self.dropout4(x)
        #print(x.size(0))
        x = x.view(batch_size, -1)
        x = self.fc1(x)
        #x = self.fc2(x)
        return x



model = Conv1DNet()






model.load_state_dict(torch.load('model1.pt',map_location=torch.device('cpu')))


# In[15]:


model.eval()




import librosa



def asr(input):
    samples, sample_rate = librosa.load(input, sr = 8000)
    if len(samples) > 8000:
        samples = audio[:8000]
    if len(samples) < 8000:
        samples = np.pad(samples, (0, 8000 - len(samples)), mode='constant')
    x_tensor=torch.from_numpy(samples)
    inputs= x_tensor
    inputs=torch.unsqueeze(inputs,0)
    y=model(inputs)
    predicted=torch.argmax(y.data)
    labels=['down', 'go', 'left', 'no', 'off', 'on', 'right', 'stop', 'up', 'yes']
    return labels[predicted]
    
    




gr.Interface(
    fn=asr, 
    inputs=gr.inputs.Audio(source="upload", type="filepath"),
    outputs="text",
    

    description="asr ",
    theme="default",

).launch()


# In[ ]: