Vodex cuts AI voice agent call drops 50% with Krisp Noise Cancellation

Customer story · Technology & Software

Vodex improves AI voice agents’ turn-taking with Krisp AI Noise Cancellation

Vodex, a Bangalore provider of generative AI voice agents for outbound calls, integrated Krisp's server-based AI Noise Cancellation to stop customer-side noise from interrupting its bots, cutting call drops 50% and lifting MOS 75%.

  • Originally published by Krisp · January 2025
  • Posted here September 13, 2026
  • 3 min read
Published by Krisp
Vodex case study
90%noise reduction
75%increase in MOS
50%fewer call drops
30%increase in CSAT

About Vodex

Vodex is a Bangalore, India-based company specializing in generative AI-powered voice agents that hold natural, humanlike conversations with customers. Its virtual agents manage the initial phases of customer interactions, giving businesses a scalable and efficient way to handle outbound sales and service calls. By personalizing conversations and providing real-time insights, Vodex helps businesses improve engagement and streamline processes.

Vodex case study

The challenge

Vodex's AI-powered agents were remarkable in their conversational capabilities, but they struggled with an all-too-common problem: noise. Background sound from the customer's side, such as traffic, household chatter or industrial noise, disrupted the bots in two ways. Ambient noise was often misinterpreted as human speech, leading to unnecessary interruptions in the conversation. The noise also reduced clarity, forcing repeated exchanges and diminishing efficiency.

Vodex explored existing options, including automated speech recognition (ASR) tools, but found them inadequate. These tools introduced delays and struggled to operate effectively in real-time scenarios, which undermined the natural turn-taking that makes an AI voice agent feel humanlike. Before Krisp, Vodex's Mean Opinion Score (MOS), a measure of perceived call quality, sat below 30.

To solve the problem, Vodex needed a way to eliminate background noise while preserving the bots' quick response times and natural conversational flow.

Krisp has transformed our product and the way we approach conversational AI. The improvement in call quality and customer satisfaction is undeniable—it's like a new chapter in what AI can achieve for sales and service.

Kumar Saurav, CTO, Vodex

The solution

Vodex partnered with Krisp, which the case study describes as the pioneer and market leader in AI-powered noise cancellation technology. Krisp's server-based noise cancellation gave Vodex a reliable way to filter disruptive background noise out of customer calls, so its bots could focus solely on the customer's voice.

By isolating human speech from background sounds, Krisp enabled Vodex's AI voice agents to process spoken responses seamlessly, without interruptions caused by noise misinterpretation, and to deliver more efficient, humanlike interactions free from awkward pauses or missed cues. The result was better turn-taking, with bots free from the awkward pauses and missed cues that noise had caused.

The results were immediate. "When our development team demonstrated Krisp's capabilities, we were blown away. Seeing our bot continue uninterrupted, even amidst loud office noise, was a gamechanger for us. It felt like a whole new level of innovation," said Kumar Saurav, CTO of Vodex.

Vodex plans to extend the collaboration, integrating Krisp into other product areas where clear audio is critical, such as improving transcription accuracy and strengthening AI voice agent performance in noisy environments. It also aims to pursue new opportunities in industries such as mortgage, insurance and contact centres/BPOs, where it expects Krisp's technology to deliver substantial value. Krisp delivers these capabilities through its Voice AI SDKs, which cancel outgoing background noise and voices, remove noise from the customer's end, localize accents in real time and offer call transcription and summarization.

The results

  • 90% noise reduction, enabling uninterrupted, clear conversations between bots and customers
  • 75% increase in Mean Opinion Score (MOS), rising from below 30 to over 75
  • 50% decrease in call drops and 15% shorter call duration
  • 20% increase in lead qualification rates and 30% increase in CSAT
  • More accurate escalations, so human agents handle only complex cases
  • One debt collection client saw 3x better recovery rates and 7x higher connect rates

Get the full case study

Download the original document as published by Krisp (PDF, 526.5 KB).

Download the case study

Your story could be next

Want results like these?

Every result on this page started with one conversation — let's have yours.