app.py

from speechbrain.pretrained.interfaces import foreign_class
import gradio as gr
import os
import warnings

warnings.filterwarnings("ignore")

# Loading the speechbrain emotion detection model
learner = foreign_class(
    source="speechbrain/emotion-recognition-wav2vec2-IEMOCAP",
    pymodule_file="custom_interface.py",
    classname="CustomEncoderWav2vec2Classifier"
)

# Emotion dictionary
emotion_dict = {
    'sad': 'Sad',
    'hap': 'Happy',
    'ang': 'Anger',
    'fea': 'Fear',
    'sur': 'Surprised',
    'neu': 'Neutral'
}

# Function for classification of uploaded files
def predict_emotion_upload(audio):
    out_prob, score, index, text_lab = learner.classify_file(audio.name)
    return emotion_dict[text_lab[0]]

# Function for classification of selected files from the dropdown
def predict_emotion_select(filename):
    file_path = os.path.join('rec', filename)
    out_prob, score, index, text_lab = learner.classify_file(file_path)
    return emotion_dict[text_lab[0]]

# Function to create an audio player component
def create_audio_player(filename):
    file_path = os.path.join('rec', filename)
    return file_path

# Retrieve a list of audio file names from the 'rec' directory
audio_files = os.listdir('rec')
audio_files_dropdown = gr.inputs.Dropdown(choices=audio_files, label="Select Audio File")

# Define Gradio interface components for both tabs
with gr.Blocks() as demo:
    gr.Markdown("## ML Speech Emotion Detection")
    gr.Markdown("Speechbrain powered wav2vec 2.0 pretrained model on IEMOCAP dataset.")

    with gr.Tabs():
        with gr.TabItem("Upload Audio"):
            with gr.Group():
                audio_upload = gr.Audio(label="Upload Audio", type="file")
                submit_btn_1 = gr.Button("Classify Uploaded Audio")
            audio_player_1 = gr.Audio(label="Uploaded Audio Player", interactive=True)
            output_text_1 = gr.Textbox(label="Prediction")

            submit_btn_1.click(predict_emotion_upload, inputs=audio_upload, outputs=[output_text_1, audio_player_1])

        with gr.TabItem("Select from List"):
            with gr.Group():
                submit_btn_2 = gr.Button("Classify Selected Audio")
            audio_player_2 = gr.Audio(label="Selected Audio Player", interactive=True)
            output_text_2 = gr.Textbox(label="Prediction")

            audio_files_dropdown.change(create_audio_player, inputs=audio_files_dropdown, outputs=audio_player_2)
            submit_btn_2.click(predict_emotion_select, inputs=audio_files_dropdown, outputs=output_text_2)

demo.launch()