Spaces:
Running
Running
import gradio as gr | |
import math | |
import time | |
def stream(audio, chunk_length_s): | |
start_time = time.time() | |
sampling_rate, array = audio | |
chunk_length = int(chunk_length_s * sampling_rate) | |
time_length = chunk_length_s / 2 # always stream outputs faster than it takes to process | |
audio_length = len(array) | |
num_batches = math.ceil(audio_length / chunk_length) | |
for idx in range(num_batches): | |
time.sleep(time_length) | |
start_pos = idx * chunk_length | |
end_pos = min((idx + 1) * chunk_length, audio_length) | |
chunk = array[start_pos : end_pos] | |
if idx == 0: | |
first_time = round(time.time() - start_time, 2) | |
run_time = round(time.time() - start_time, 2) | |
yield (sampling_rate, chunk), first_time, run_time | |
with gr.Blocks() as demo: | |
with gr.Row(): | |
with gr.Column(): | |
audio_in = gr.Audio(value="librispeech.wav", sources=["upload"], type="numpy") | |
chunk_length = gr.Slider(minimum=1, maximum=3, value=2, step=1, label="Chunk length (s)") | |
run_button = gr.Button("Stream audio") | |
with gr.Column(): | |
audio_out = gr.Audio(streaming=True, autoplay=True) | |
first_time = gr.Textbox(label="Time to first chunk (s)") | |
run_time = gr.Textbox(label="Time to current chunk (s)") | |
run_button.click(fn=stream, inputs=[audio_in, chunk_length], outputs=[audio_out, first_time, run_time]) | |
demo.launch() |