QuickLabeler / app.py
msis's picture
reduce verbosity of nemo and print output
113d436
raw
history blame
2.13 kB
import os
import gradio as gr
import torch
import nemo.collections.asr as nemo_asr
import wandb
from pydub.utils import mediainfo
MODEL_HISTORY_DAYS = 180
WANDB_ENTITY = os.environ.get("WANDB_ENTITY", "tarteel")
WANDB_PROJECT_NAME = os.environ.get("WANDB_PROJECT_NAME", "nemo-experiments")
MODEL_NAME = os.environ.get("MODEL_NAME", "CfCtcLg-SpeUni1024-DI-EATLDN-CA:v0")
def get_device():
if torch.cuda.is_available():
return "cuda"
else:
return "cpu"
# run = wandb.init(entity=WANDB_ENTITY, project=WANDB_PROJECT_NAME)
wandb_api = wandb.Api(overrides={"entity": WANDB_ENTITY})
artifact = wandb_api.artifact(f"{WANDB_ENTITY}/{WANDB_PROJECT_NAME}/{MODEL_NAME}")
artifact_dir = artifact.download()
# find the model (ending with .nemo) in the artifact directory
model_path = [
os.path.join(root, file)
for root, dirs, files in os.walk(artifact_dir)
for file in files
if file.endswith(".nemo")
][0]
model = nemo_asr.models.EncDecCTCModelBPE.restore_from(
model_path, map_location=get_device()
)
def transcribe(audio_file):
transcription = model.transcribe([audio_file], verbose=False)[0]
print(f"{audio_file}: {transcription}")
return transcription
def get_duration_ms(audio_file):
duration = mediainfo(audio_file)["duration"] # a string in seconds
duration_ms = int(float(duration) * 1000)
print(f"{audio_file}: {duration_ms} ms")
return duration_ms
demo = gr.Blocks()
with demo:
gr.Markdown(
"""
# ﷽
"""
)
with gr.Row():
audio_file = gr.Audio(source="upload", type="filepath", label="File")
with gr.Row():
output_file = gr.TextArea(label="Audio Transcription")
b1 = gr.Button("Transcribe")
b1.click(
transcribe,
inputs=[audio_file],
outputs=[output_file],
api_name="transcribe",
)
b2 = gr.Button("Get Duration")
with gr.Row():
duration = gr.TextArea(label="Duration")
b2.click(
get_duration_ms,
inputs=[audio_file],
outputs=[duration],
api_name="get_duration_ms",
)
demo.launch()