Extracted voice is not clear 

 

1. I would like to create an offline app for speech recognition (speech to text) in a noisy environment (recording voice in high-noise area).
However, after noise extraction using spleeter, the speech-to-text conversion is not accurate because the quality of the extracted voice is low.

2. What is the maximum value of noise (in Hz) that supports the noise extraction in spleeter

I am attaching the code below.

**from spleeter.separator import Separator
from spleeter.audio.adapter import AudioAdapter
from scipy.io.wavfile import write
import numpy as np

separator = Separator('spleeter:2stems')
audio_loader = AudioAdapter.default()
sampleRate = 41050
waveform, _ = audio_loader.load("VoiceSample_HighNoise.wav", sample_rate=sr)
prediction = separator.separate(waveform)
write("VoiceSample_HighNoise_Output.wav", sampleRate, prediction['vocals'].astype(np.float32))**


Is there any need to change the above code for extracting the High Clarity in voice.