How to Remove Background Noise for STT: Complete Guide to Noise Reduction for Speech-to-Text

Why Remove Background Noise for STT?

• Reduced signal-to-noise ratio (SNR) makes it harder for models to distinguish speech
• Frequency masking where noise overlaps with speech frequencies
• Model confusion when noise patterns resemble speech
• Lower confidence scores leading to more transcription errors
• Increased processing time as models struggle with noisy input

• ✅ Improved transcription accuracy (often 10-30% better)
• ✅ Better word recognition, especially for technical terms
• ✅ Faster processing with cleaner audio
• ✅ More reliable timestamps and segmentation
• ✅ Better handling of quiet speech

Understanding Background Noise Types

1. Constant Noise (Stationary)

• Examples: Air conditioning, fan hum, electrical hum, white noise
• Characteristics: Consistent frequency and amplitude
• Removal: Easier to remove with spectral subtraction or filtering

2. Variable Noise (Non-Stationary)

• Examples: Traffic, crowd chatter, keyboard typing, paper rustling
• Characteristics: Changes over time, unpredictable patterns
• Removal: Requires more advanced techniques like deep learning models

3. Impulsive Noise

• Examples: Clicks, pops, door slams, phone rings
• Characteristics: Short, sudden bursts
• Removal: Requires detection and replacement/interpolation

4. Periodic Noise

• Examples: Beeping, alarms, repetitive sounds
• Characteristics: Regular patterns at specific frequencies
• Removal: Can be filtered with notch filters

Method 1: Using Audio Editing Software

Audacity (Free, Open Source)

1. Open your audio file in Audacity
2. Select a section with only noise (no speech)
3. Go to Effect → Noise Reduction
4. Click Get Noise Profile
5. Select the entire audio track
6. Go to Effect → Noise Reduction again
7. Adjust settings:
   • Noise reduction (dB): 12-24 dB (start with 15)
   • Sensitivity: 6.0 (default)
   • Frequency smoothing (bands): 3 (default)
8. Click OK to apply

• Use a noise sample of 0.5-2 seconds
• Choose a section with representative noise
• Start with moderate settings and increase if needed
• Preview before applying to full track

Adobe Audition

1. Open audio file
2. Select noise-only section
3. Go to Effects → Noise Reduction/Restoration → Capture Noise Print
4. Select entire track
5. Go to Effects → Noise Reduction/Restoration → Noise Reduction (process)
6. Adjust:
   • Noise Reduction: 40-80% (start with 60%)
   • Reduce by: 6-12 dB
   • High Frequency Transition: 4000-8000 Hz
7. Click Apply

Method 2: Python Audio Processing Libraries

Using noisereduce Library

import noisereduce as nr
import soundfile as sf

# Load audio file
audio_data, sample_rate = sf.read("noisy_audio.wav")

# Method 1: Stationary noise reduction (for constant noise)
reduced_noise = nr.reduce_noise(
    y=audio_data,
    sr=sample_rate,
    stationary=True,
    prop_decrease=0.8  # Reduce noise by 80%
)

# Method 2: Non-stationary noise reduction (for variable noise)
reduced_noise = nr.reduce_noise(
    y=audio_data,
    sr=sample_rate,
    stationary=False,
    prop_decrease=0.8
)

# Save cleaned audio
sf.write("cleaned_audio.wav", reduced_noise, sample_rate)

pip install noisereduce soundfile

Using librosa for Spectral Gating

import librosa
import numpy as np
import soundfile as sf

def spectral_gate(audio_path, threshold_db=-40):
    """Remove noise using spectral gating."""
    # Load audio
    y, sr = librosa.load(audio_path, sr=None)
    
    # Compute short-time Fourier transform (STFT)
    stft = librosa.stft(y)
    magnitude = np.abs(stft)
    phase = np.angle(stft)
    
    # Convert to dB
    magnitude_db = librosa.amplitude_to_db(magnitude)
    
    # Apply threshold (remove frequencies below threshold)
    magnitude_db_cleaned = np.where(
        magnitude_db > threshold_db,
        magnitude_db,
        -80  # Silence very quiet parts
    )
    
    # Convert back to linear scale
    magnitude_cleaned = librosa.db_to_amplitude(magnitude_db_cleaned)
    
    # Reconstruct audio
    stft_cleaned = magnitude_cleaned * np.exp(1j * phase)
    y_cleaned = librosa.istft(stft_cleaned)
    
    return y_cleaned, sr

# Usage
cleaned_audio, sample_rate = spectral_gate("noisy_audio.wav", threshold_db=-35)
sf.write("cleaned_audio.wav", cleaned_audio, sample_rate)

Using scipy for High-Pass Filtering

from scipy import signal
import soundfile as sf

def high_pass_filter(audio_path, cutoff_freq=80):
    """Remove low-frequency noise with high-pass filter."""
    # Load audio
    audio_data, sample_rate = sf.read(audio_path)
    
    # Design high-pass filter
    nyquist = sample_rate / 2
    normalized_cutoff = cutoff_freq / nyquist
    b, a = signal.butter(4, normalized_cutoff, btype='high')
    
    # Apply filter
    filtered_audio = signal.filtfilt(b, a, audio_data)
    
    return filtered_audio, sample_rate

# Usage
cleaned_audio, sr = high_pass_filter("noisy_audio.wav", cutoff_freq=100)
sf.write("cleaned_audio.wav", cleaned_audio, sr)

Method 3: Deep Learning-Based Noise Reduction

Using RNNoise

import rnnoise
import numpy as np
import soundfile as sf

def rnnoise_denoise(audio_path):
    """Remove noise using RNNoise model."""
    # Load audio
    audio_data, sample_rate = sf.read(audio_path)
    
    # RNNoise expects 16kHz mono audio
    if sample_rate != 16000:
        import librosa
        audio_data = librosa.resample(audio_data, orig_sr=sample_rate, target_sr=16000)
        sample_rate = 16000
    
    # Convert to mono if stereo
    if len(audio_data.shape) > 1:
        audio_data = np.mean(audio_data, axis=1)
    
    # Process in chunks (RNNoise processes 480 samples at a time)
    chunk_size = 480
    denoised_audio = []
    
    denoiser = rnnoise.RNNoise()
    
    for i in range(0, len(audio_data), chunk_size):
        chunk = audio_data[i:i+chunk_size]
        if len(chunk) < chunk_size:
            chunk = np.pad(chunk, (0, chunk_size - len(chunk)))
        
        denoised_chunk = denoiser.process(chunk)
        denoised_audio.extend(denoised_chunk)
    
    return np.array(denoised_audio), sample_rate

# Usage
cleaned_audio, sr = rnnoise_denoise("noisy_audio.wav")
sf.write("cleaned_audio.wav", cleaned_audio, sr)

pip install rnnoise

Using Facebook's Demucs

from demucs.pretrained import get_model
from demucs.audio import AudioFile
import torch

def demucs_separation(audio_path):
    """Separate speech from noise using Demucs."""
    # Load pre-trained model
    model = get_model('htdemucs')
    model.eval()
    
    # Load audio
    wav = AudioFile(audio_path).read(streams=0, samplerate=model.sample_rate, channels=model.audio_channels)
    ref = wav.mean(0)
    wav = (wav - ref.mean()) / ref.std()
    wav = torch.from_numpy(wav).float()
    
    # Separate sources
    with torch.no_grad():
        sources = model(wav[None])
        sources = sources * ref.std() + ref.mean()
    
    # Extract vocals (speech) - usually index 0 or 3
    speech = sources[0, 0].cpu().numpy()
    
    return speech, model.sample_rate

# Usage
speech_audio, sr = demucs_separation("noisy_audio.wav")
sf.write("speech_only.wav", speech_audio, sr)

Method 4: Online Noise Reduction Tools

1. Audacity Online (Cloud Version)

• Free, browser-based
• Good for quick processing
• Limited file size

2. Adobe Podcast Enhance

• AI-powered noise reduction
• Free for limited use
• Excellent results for speech

3. Krisp.ai

• Real-time noise suppression
• API available for integration
• Good for live audio

4. Cleanvoice.ai

• Automatic noise removal
• Handles multiple noise types
• Batch processing available

Complete Workflow: Preprocessing Audio for STT

import librosa
import noisereduce as nr
import soundfile as sf
from scipy import signal
import numpy as np

def preprocess_audio_for_stt(audio_path, output_path):
    """Complete audio preprocessing pipeline for STT."""
    
    # Step 1: Load audio
    print("Loading audio...")
    y, sr = librosa.load(audio_path, sr=16000, mono=True)
    
    # Step 2: Remove DC offset
    print("Removing DC offset...")
    y = y - np.mean(y)
    
    # Step 3: High-pass filter (remove low-frequency noise)
    print("Applying high-pass filter...")
    nyquist = sr / 2
    normalized_cutoff = 80 / nyquist
    b, a = signal.butter(4, normalized_cutoff, btype='high')
    y = signal.filtfilt(b, a, y)
    
    # Step 4: Normalize volume
    print("Normalizing volume...")
    max_val = np.max(np.abs(y))
    if max_val > 0:
        y = y / max_val * 0.95  # Normalize to 95% to avoid clipping
    
    # Step 5: Noise reduction
    print("Reducing noise...")
    y = nr.reduce_noise(
        y=y,
        sr=sr,
        stationary=False,  # Use non-stationary for variable noise
        prop_decrease=0.8  # Reduce noise by 80%
    )
    
    # Step 6: Final normalization
    print("Final normalization...")
    max_val = np.max(np.abs(y))
    if max_val > 0:
        y = y / max_val * 0.95
    
    # Step 7: Save processed audio
    print(f"Saving to {output_path}...")
    sf.write(output_path, y, sr)
    
    print("Preprocessing complete!")
    return y, sr

# Usage
preprocess_audio_for_stt("noisy_recording.wav", "cleaned_for_stt.wav")

Best Practices for Noise Removal

1. Choose the Right Method

• Constant noise: Use spectral subtraction or stationary noise reduction
• Variable noise: Use non-stationary reduction or deep learning models
• Impulsive noise: Use click removal or interpolation
• Multiple noise types: Combine multiple techniques

2. Preserve Speech Quality

• Don't over-process (can introduce artifacts)
• Use moderate noise reduction settings (60-80%)
• Preserve frequency range of human speech (80-8000 Hz)
• Maintain natural speech characteristics

3. Test and Iterate

• Always preview before applying to full track
• Compare original vs. processed audio
• Test transcription accuracy with both versions
• Adjust settings based on results

4. Consider Your STT Model

• Some models (like Whisper) handle noise well
• Preprocessing may not always be necessary
• Test with and without preprocessing
• Larger models are more noise-robust

Common Mistakes to Avoid

• Can remove speech frequencies
• Creates artifacts and distortion
• Makes speech sound robotic

• Can remove important speech components
• Makes speech sound thin or tinny
• Affects naturalness

• Preprocessing may not improve accuracy
• Some models work better with original audio
• Always A/B test

• Ensure proper sample rate (16kHz recommended)
• Use lossless formats when possible
• Avoid double compression

Integration with Speech-to-Text

Using with OpenAI Whisper

import whisper
import noisereduce as nr
import soundfile as sf

def transcribe_with_noise_reduction(audio_path):
    """Transcribe audio with noise reduction preprocessing."""
    
    # Step 1: Reduce noise
    audio_data, sr = sf.read(audio_path)
    cleaned_audio = nr.reduce_noise(
        y=audio_data,
        sr=sr,
        stationary=False,
        prop_decrease=0.75
    )
    
    # Save temporary cleaned file
    temp_path = "temp_cleaned.wav"
    sf.write(temp_path, cleaned_audio, sr)
    
    # Step 2: Transcribe with Whisper
    model = whisper.load_model("base")
    result = model.transcribe(temp_path)
    
    # Clean up
    import os
    os.remove(temp_path)
    
    return result["text"]

# Usage
transcription = transcribe_with_noise_reduction("noisy_audio.wav")
print(transcription)

Using with SayToWords API

import requests
import noisereduce as nr
import soundfile as sf

def transcribe_with_saytowords(audio_path):
    """Preprocess and transcribe with SayToWords."""
    
    # Preprocess audio
    audio_data, sr = sf.read(audio_path)
    cleaned_audio = nr.reduce_noise(
        y=audio_data,
        sr=sr,
        stationary=False,
        prop_decrease=0.8
    )
    
    # Save cleaned audio
    cleaned_path = "cleaned_for_api.wav"
    sf.write(cleaned_path, cleaned_audio, sr)
    
    # Upload and transcribe
    with open(cleaned_path, 'rb') as f:
        files = {'file': f}
        response = requests.post(
            'https://api.saytowords.com/transcribe',
            files=files,
            headers={'Authorization': 'Bearer YOUR_API_KEY'}
        )
    
    return response.json()

Measuring Noise Reduction Effectiveness

Before/After Comparison

import librosa
import numpy as np

def measure_snr(audio_path):
    """Estimate signal-to-noise ratio."""
    y, sr = librosa.load(audio_path, sr=None)
    
    # Simple SNR estimation
    signal_power = np.mean(y ** 2)
    noise_floor = np.percentile(np.abs(y), 10) ** 2
    snr_db = 10 * np.log10(signal_power / noise_floor) if noise_floor > 0 else 0
    
    return snr_db

# Compare before and after
original_snr = measure_snr("noisy_audio.wav")
cleaned_snr = measure_snr("cleaned_audio.wav")

print(f"Original SNR: {original_snr:.2f} dB")
print(f"Cleaned SNR: {cleaned_snr:.2f} dB")
print(f"Improvement: {cleaned_snr - original_snr:.2f} dB")

Conclusion

• Noise type (constant vs. variable)
• Audio quality (sample rate, bit depth)
• Available tools (software vs. programming)
• STT model (some handle noise better than others)

• For quick processing: Use Audacity or online tools
• For automation: Use Python libraries like noisereduce
• For best results: Combine multiple techniques
• For production: Test with your specific STT model

Additional Resources

• Whisper for Noisy Background Guide
• How to Fix Unclear Recordings
• Audio Enhancement Best Practices