How does V100's noise cancellation work?

V100's noise suppression uses a Web Audio API pipeline that chains three processing stages: a high-pass filter that removes low-frequency noise (HVAC, traffic rumble, fans), a dynamic range compressor that normalizes volume levels and reduces sudden noise spikes, and a limiter that prevents audio clipping. The pipeline processes audio in real time with under 10ms added latency. Developers control the suppression level (low, medium, high) through the API.

How does V100's noise suppression compare to Krisp?

Krisp uses a deep learning model for noise suppression that is exceptionally effective at removing complex noise like barking dogs, construction, and overlapping conversations. V100's approach uses a Web Audio signal processing pipeline that is lighter weight and runs natively in the browser without a separate model download. Krisp produces better results on challenging noise but requires a desktop app or SDK integration. V100's noise suppression is API-controlled and runs entirely in the browser with zero additional downloads.

Can I control noise suppression per participant?

Yes. V100's noise suppression can be enabled, disabled, or adjusted per participant through the REST API or client SDK. You can set different suppression levels for different participants in the same session. For example, a speaker in a quiet studio might use low suppression while a participant in a coffee shop uses high suppression.

NOISE SUPPRESSION

Video API with
Noise Cancellation

Background noise is the number one complaint in video calls. Keyboard clatter, HVAC hum, barking dogs, and street traffic degrade audio quality and make transcription unreliable. V100's noise suppression API uses a Web Audio processing pipeline -- high-pass filter, dynamic compressor, and limiter -- to clean audio in real time with under 10ms of added latency. Enable it with a single API parameter. Adjust the suppression level per participant. No additional software or model downloads required.

Get API Key View Code Sample

Under 10ms added latency

3 suppression levels

One-line API toggle

The Background Noise Problem

Remote work moved meetings out of sound-treated conference rooms and into kitchens, coffee shops, and shared apartments. The audio quality gap between a controlled office environment and a typical home setup is significant, and it has downstream effects on transcription accuracy, meeting productivity, and participant experience.

Transcription Accuracy Drops

Speech-to-text engines perform significantly worse in noisy environments. A keyboard typing over a speaker's voice causes word-level errors that propagate into meeting summaries, action items, and searchable transcripts. Noise suppression before the transcription engine improves accuracy by 15-30% in typical home office environments.

Cognitive Load Increases

Participants in noisy environments spend mental energy filtering background sounds instead of focusing on the conversation. Research from the University of London shows that background noise in meetings reduces information retention by up to 25%. Clean audio is not a luxury -- it is a productivity requirement.

Professional Impression Suffers

In sales calls, client presentations, and legal consultations, background noise signals unprofessionalism. A barking dog during a sales demo has a measurable negative impact on deal progression. Noise suppression is a professional requirement for customer-facing video, not an optional enhancement.

The Web Audio Processing Pipeline

V100's noise suppression chains three Web Audio API nodes into a real-time processing pipeline. Each stage targets a different class of noise. The pipeline runs entirely in the browser with no server round-trip, adding under 10ms of latency to the audio path.

Stage 1: High-Pass Filter

A BiquadFilterNode configured as a high-pass filter removes low-frequency noise below a configurable cutoff frequency. This eliminates HVAC rumble, traffic drone, washing machines, refrigerator hum, and other constant low-frequency sounds that occupy the 20-300Hz range. The human voice fundamental frequency starts around 85Hz for men and 165Hz for women, so a cutoff around 80-100Hz removes most environmental noise without affecting speech intelligibility.

Low: 80Hz cutoff | Medium: 120Hz | High: 200Hz

Stage 2: Dynamic Compressor

A DynamicsCompressorNode reduces the dynamic range of the audio signal. This normalizes volume differences between a soft-spoken participant and a loud one, reduces sudden noise spikes (keyboard clicks, door slams, notification sounds), and creates a more consistent listening experience. The compressor's threshold, knee, ratio, attack, and release parameters are tuned per suppression level.

Low: -24dB threshold, 4:1 | Medium: -30dB, 8:1 | High: -40dB, 12:1

Stage 3: Limiter

A second DynamicsCompressorNode configured as a brick-wall limiter prevents the output signal from exceeding 0dBFS. This eliminates audio clipping that causes distortion and protects downstream systems (speakers, recording codecs, transcription engines) from signal peaks. The limiter acts as a safety net -- it should rarely activate when the compressor is doing its job, but when it does, it prevents the worst-case audio artifacts.

Threshold: -1dBFS | Ratio: 20:1 | Attack: 0.001s

Three Suppression Levels

Not all environments need the same level of noise suppression. Aggressive suppression in a quiet room can make speech sound unnatural. Light suppression in a noisy cafe leaves too much noise. V100 exposes three levels so you can match the suppression to the environment.

LOW

Quiet Environments

Light filtering for quiet home offices and conference rooms. Removes HVAC hum and subtle fan noise without affecting voice quality. Preserves the most natural sound. Best for professional recording environments where audio quality is the priority.

80Hz high-pass cutoff

Gentle compression (4:1 ratio)

Most natural voice reproduction

MEDIUM (DEFAULT)

Home Office

Balanced filtering for typical work-from-home environments. Handles keyboard typing, household appliances, light traffic noise, and casual conversation in adjacent rooms. The default level for most use cases -- effective noise reduction without making speech sound processed.

120Hz high-pass cutoff

Moderate compression (8:1 ratio)

Best balance of quality and suppression

HIGH

Noisy Environments

Aggressive filtering for coffee shops, coworking spaces, and outdoor environments. Removes most background noise including nearby conversations, music, and traffic. Voice may sound slightly compressed, but intelligibility is preserved in environments that would otherwise be unusable for video calls.

200Hz high-pass cutoff

Heavy compression (12:1 ratio)

Maximum noise reduction

Comparison: V100 vs Krisp vs Zoom vs Teams

Different products take different approaches to noise suppression. Here is an honest comparison of how V100's signal processing pipeline stacks up against deep learning approaches and built-in platform noise suppression.

Capability	V100	Krisp	Zoom	Teams
Approach	Web Audio DSP	Deep learning	Proprietary ML	Proprietary ML
API control	Full REST API	SDK only	User UI	User/Admin UI
Complex noise removal	Good	Excellent	Very good	Very good
Per-participant control	Yes	No	No	No
Additional download	None	App or SDK	None (built-in)	None (built-in)
Added latency	<10ms	~20ms	Not published	Not published

Honest trade-off: Krisp's deep learning model is genuinely better at removing complex non-stationary noise like dog barks, baby cries, and overlapping conversations. V100's Web Audio pipeline excels at stationary noise (fans, HVAC, traffic) and transient noise (keyboard clicks, notification sounds). If your users regularly work in extremely noisy environments, Krisp's specialized approach will outperform signal processing. V100's advantage is zero-download browser-native processing with full API control.

Integration Guide

Enable noise suppression with a single parameter when creating a session. Adjust levels per participant during the meeting. Toggle on or off at any time through the REST API.

noise-suppression.js

// npm install v100-sdk
import { V100 } from 'v100-sdk';

const v100 = new V100('YOUR_API_KEY');

// Create a session with noise suppression enabled
const session = await v100.meetings.create({
  name: 'Client Strategy Call',
  noiseSuppression: {
    enabled: true,
    level: 'medium',           // 'low' | 'medium' | 'high'
    applyToAll: true            // Default level for all participants
  },
  transcription: {
    enabled: true               // Cleaner audio = better transcription
  }
});

// Adjust a specific participant's suppression level
await v100.meetings.updateParticipant(session.id, 'participant_abc', {
  noiseSuppression: {
    level: 'high'              // This person is in a coffee shop
  }
});

// Toggle noise suppression off for a participant
await v100.meetings.updateParticipant(session.id, 'participant_xyz', {
  noiseSuppression: {
    enabled: false             // They're in a quiet studio
  }
});

Or use the REST API directly:

terminal

curl -X POST https://api.v100.ai/v1/meetings \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Weekly Standup",
    "noiseSuppression": {
      "enabled": true,
      "level": "medium"
    },
    "transcription": { "enabled": true }
  }'

Pricing

Noise suppression is included on all V100 plans at no additional cost. Since the Web Audio processing runs on-device in the browser, it does not consume V100 server resources and incurs no additional billing.

Free Tier

$0/mo

All suppression levels included.

3 suppression levels

API toggle

Per-participant control

60 minutes/month

RECOMMENDED

Pro

Usage-based

Production noise suppression at scale.

Everything in Free

Unlimited minutes

Audio quality analytics

Priority support

Enterprise

Custom

Custom audio processing pipelines.

Everything in Pro

Custom filter chains

Audio processing webhooks

Dedicated support

Add Noise Suppression to Your Video App Today

Free tier includes all three suppression levels and per-participant control. One API parameter to enable. Under 10ms of added latency. No additional downloads.

Get API Key — Free Tier Virtual Backgrounds

Related Resources

Feature

Video API with
Noise Cancellation

The Background Noise Problem

Transcription Accuracy Drops

Cognitive Load Increases

Professional Impression Suffers

The Web Audio Processing Pipeline

Stage 1: High-Pass Filter

Stage 2: Dynamic Compressor

Stage 3: Limiter

Three Suppression Levels

Quiet Environments

Home Office

Noisy Environments

Comparison: V100 vs Krisp vs Zoom vs Teams

Integration Guide

Pricing

Add Noise Suppression to Your Video App Today

Related Resources

Virtual Backgrounds API

Speaker Diarization API

Transcription API

Video API withNoise Cancellation

The Background Noise Problem

Transcription Accuracy Drops

Cognitive Load Increases

Professional Impression Suffers

The Web Audio Processing Pipeline

Stage 1: High-Pass Filter

Stage 2: Dynamic Compressor

Stage 3: Limiter

Three Suppression Levels

Quiet Environments

Home Office

Noisy Environments

Comparison: V100 vs Krisp vs Zoom vs Teams

Integration Guide

Pricing

Add Noise Suppression to Your Video App Today

Related Resources

Virtual Backgrounds API

Speaker Diarization API

Transcription API

Video API with
Noise Cancellation