Background Denoising
Background denoising helps Vomyra understand callers in real-world audio: traffic, fans, TV, office noise, storefront chatter, and other non-caller sounds that can pollute transcripts or trigger false interruptions.
Clean audio & debugging
Most deployments
Very noisy environments
What it does
Clean caller audio before transcription
Background denoising filters the caller audio stream before Vomyra sends it to speech-to-text. It helps the agent focus on the caller's voice instead of fans, traffic, TV, keyboard noise, shop sounds, or other people speaking nearby.
Cleaner transcript
Less background noise reaches STT, so transcripts contain fewer accidental words.
Fewer false interrupts
Noise is less likely to make the agent stop speaking or answer too early.
Better live calls
The agent behaves more consistently in offices, shops, homes, and mobile environments.
Note
Before and after
What changes when denoising is enabled
The practical effect is visible in the transcript. Denoising removes non-caller audio before speech recognition, so the language model receives cleaner text.
Without Denoising
With Denoising
Accuracy impact
Impact on speech recognition
Background denoising directly affects the audio sent to the speech recognizer. It can improve transcript quality, but very strong filtering should be tested carefully.
Benefits
Cleaner transcripts
Reduced false interruptions
Better performance in noisy environments
Potential trade-offs
Very aggressive filtering may remove portions of speech
Short responses may become harder to detect
Multilingual accuracy can occasionally decrease
When to enable
Use it when the caller's environment is unpredictable
Most production phone agents should start with Standard denoising. Turn it off only when the audio is already clean or when you are debugging missing short utterances.