Skip to main content
Improve ASR accuracy for conversational AI in noisy environments

​
Overview

Voice AI agents struggle with speech recognition in real-world environments. Background noise, chatter, and ambient sounds increase Word Error Rates (WER), leading to misunderstandings and poor user experience. The AI Voice Agent model solves this by applying specialized noise cancellation optimized for ASR pipelines—preserving acoustic features that speech recognition models need while removing distracting noise. Result: 5-30% WER improvement without degradation on clean audio.

​
How It Works

1

Take input from the audio pipeline

Deployed in the existing audio pipeline for the user’s speech.
2

Process through Sanas AI Voice Agent

Model removes background noise while preserving phonetic information for ASR.
3

Send clean audio to your ASR

Works with any ASR: Deepgram, Google, Whisper, Azure, etc.
4

Get an accurate transcription

Lower WER means better downstream NLP and conversation quality.

​
Key Benefits

5-30% WER Improvement

Proven reduction in Word Error Rate across major ASR systems.

ASR-Agnostic

Works with any ASR: Deepgram, Google, Whisper, Azure, AWS.

No Clean Audio Penalty

No degradation when the denoised audio is fed into ASR systems.

Real-Time Performance

Low latency suitable for live conversations.

​
Use Cases

IndustryUse CaseImpact
Customer ServiceAI phone support botsHigher first-call resolution
HealthcareMedical transcription agentsAccurate diagnosis documentation
FinanceBanking voice assistantsReduced transaction errors
AutomotiveIn-car voice commandsWorks in highway noise
Smart HomeVoice assistantsKitchen/TV noise immunity

​
Performance Example

​
Real-World Test: Background Chatter & Noise

Environment: Speaker with cutlery sounds, background conversation, restaurant ambient noise ASR Used: Deepgram Nova3 Streaming Transcription Accuracy & Audio Samples
Oracle TranscriptSource TranscriptGeneric BVCSanas AI Voice Agent
it was in the spring of the year eighteen ninety four that all london was interested and the fashionable world dismayed by the murder of the honorable ronald adair under most unusual and inexplicable circumstances the public has already learned those particulars of the crime which came out in the police investigation but a good deal was suppressed upon that occasion since the case for the prosecution was so overwhelmingly strong that it was not necessary to bring forward all the facts only now at the end of nearly ten years am i allowed to supply those missing links which make up the whole of the remarkable chain the crime was of interest in itself but that interest was as nothing to me compared to the incons inconceivable sequence which afforded me the greatest shock and surprise of any event in my adventurous lifeit was in the spring of the year one eight nine four that all london was interested and the fashionable world dismayed by the mother of the honorable rona adair under most unusual and those particulars of the crime which came out in the police investigation but a good deal was suppressed upon the occasion since the case for the prosecution strong that it was not necessary necessary to bring forward all the facts only now at the end of nearly ten years am i allowed to supplement these basic things which make the crime was of interest in itself but that interest was as nothing to me compared to the incon inconceivable sequence which afforded me the greatest shock and of any event in my adventurous life let’s create magicit was in the one eight nine four that all london was interested and the fashionable world dismayed by the mother of the honorable rona adair and the most unusual and inexplicable circumstances the public has already learned those particulars of the crime which came out in the police investigation but a good deal was suppressed upon the occasion since the case for the prosecution was so overwhelmingly strong that it was not necessary to bring forward all the facts only now at the end of nearly ten years number allowed to supply these basic wings which make up the board of the remarkable chain the crime was of interest in itself but that interest was as nothing to me compared to the incomes inconsistencyable sequence which afforded me the greatest shock and surprise of any event in my adventurous lifeit was in the spring of the year one eight nine four that all london was interested and the fashionable world dismayed by the murder of the honorable rona adair and the most unusual and inexplicable circumstances the public has already learned those particulars of the crime which came out in the police investigation but a good deal was surprised upon the occasion since the case for the prosecution was so overwhelmingly strong that it was not necessary to bring forward all the facts only now at the end of nearly ten years am i allowed to supply these missing links of which make up the whole of the remarkable chain the crime was of interest in itself but that interest was as nothing to me compared to the inconceivable sequence which afforded me the greatest shock and surprise of any event in my adventurous life
—22.9% WER13.2% WER7.5% WER

​
Quick Start (30 seconds)

import sanas_remote_sdk

# Initialize SDK
sdk = sanas_remote_sdk.CreateRemoteSDK()
init_params = sanas_remote_sdk.InitParams()
init_params.remoteEndpoint = "your_endpoint"
init_params.accountId = "your_account_id"
init_params.accountSecret = "your_account_secret"
sdk.Initialize(init_params)

# Create AI Voice Agent processor
audio_params = sanas_remote_sdk.AudioParams()
audio_params.modelName = "AGENTIC_NC"
audio_params.sampleRate = 16000

processor, result = sdk.CreateAudioProcessor(audio_params, state_callback)

# Process audio in real-time
for chunk in audio_stream:
    clean_audio = processor.ProcessSamples(chunk)
    # Send clean_audio to your ASR

Model Guide

Technical specification and parameters for the AI Voice Agent model.

Quickstart guide

Install libraries and process your first audio stream.

​
Deployment Options

Deploy the AI Voice Agent model through the following platform:
Recommended: Fastest way to production
  • Managed infrastructure
  • Auto-scaling
  • Currently available in Mumbai (AWS ap-south-1), more regions to be added
Get started →

​
Next Steps

Request Access

Get your SDK credentials

Quickstart guide

Install dependencies to get started.