---
title: "Independent benchmark: Hiya is the fastest production-ready deepfake detector"
description: Hiya scored 99.14% accuracy on the independent Podonos Deepfake Audio Detection Benchmark, with the fastest response time of the five systems above 99%.
image: https://blog.hiya.com/hubfs/Podonos%20blog%20hero.png
---

[![Hiya logo](https://blog.hiya.com/hubfs/hiya-new-purple.svg)](https://www.hiya.com/)

Solutions

Solutions

[For Mobile Operators Protect your network](https://www.hiya.com/solutions/operators) [For Businesses Brand your calls](https://www.hiya.com/solutions/businesses) [For People Get the mobile app](https://www.hiya.com/products/apps/hiya-spam-blocker)

###### Webinar

Branded Calling 101 Weekly Webinar

Make your company's calls more recognizable. Learn how Hiya can drive value for your business.

[Sign up today Arrow Right](https://www.hiya.com/lp/webinar-series-intro-to-hiya-branded-call)

Products

 Connect

###### [Branded Call Display your branded caller ID](https://www.hiya.com/products/connect/branded-call)

###### [Number Registration Display your branded caller ID](https://www.hiya.com/products/connect/number-registration)

###### [View Plans Flexible pricing for teams of all sizes](https://www.hiya.com/products/connect/pricing)

 Protect

###### [Spam Analytics Stop spam & fraud on your mobile network](https://www.hiya.com/products/protect/spam-analytics)

###### [AI Voice Detection Real-time AI voice detection](https://www.hiya.com/products/protect/ai-voice-detection)

 Mobile apps

###### [Hiya Spam Blocker Fraud & AI voice protection](https://www.hiya.com/products/apps/hiya-spam-blocker)

###### [Hiya AI Phone Productivity for busy people](https://www.hiya.com/products/apps/hiya-ai-phone)

![](https://blog.hiya.com/hubfs/sotc-report.avif)

###### Report

State of the Call 2026

86% of unidentified calls go unanswered. Read the benchmark report for what is happening in voice today, and what you can do to drive business.

[Read the report Arrow Right](https://www.hiya.com/state-of-the-call)

Why Hiya?

Overview

[Why Hiya Your voice innovation partner](https://www.hiya.com/why-hiya) [How it works Get started quickly & easily](https://www.hiya.com/how-it-works) [Customer Stories Real companies, real results](https://www.hiya.com/case-studies) [Voice Intelligence Platform Industry’s leading voice platform](https://www.hiya.com/why-hiya/voice-intelligence-platform) [Trust Center Compliance, security, & privacy](https://www.hiya.com/why-hiya/trust-center)

Company

[About Leadership and history](https://www.hiya.com/company/about) [Careers We're hiring!](https://www.hiya.com/company/careers) [Contact us Get in touch](https://www.hiya.com/contact-us)

![](https://blog.hiya.com/hubfs/bclc-hero-img.webp)

###### Customer Story

BCLC increases business KPIs with Hiya

With Hiya Branded Call BCLC was able to increase contact rates, campaign efficiency, and revenue.

[Read their story Arrow Right](https://www.hiya.com/customer-stories/bclc)

Resources

Resources

[Resource Center](https://www.hiya.com/resources) [Partner Program](https://partners.hiya.com/) [Get Support](https://hiya.com/support) [Developer Docs](https://developer.hiya.com/)

[Hiya Blog](https://blog.hiya.com/) [Newsroom](https://www.hiya.com/newsroom) [Events](https://www.hiya.com/events)

###### eBook

10 Tips for customer-friendly phone calls

 Prevent caller reputation issues and complaints with customer-friendly calling practices. 

[Read eBook Arrow Right](https://www.hiya.com/resources/ebooks/10-tips-for-customer-friendly-phone-calls)

[Get Started Arrow Right](https://www.hiya.com/#audience)

[Log in](https://connect.hiya.com/login) [Get started](https://www.hiya.com/#audience)

- [Home](https://hiya.com)
- [Blog](https://blog.hiya.com)
- [Independent benchmark: Hiya is the fastest production-ready ...](https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector)

[Deepfake Voice Detection](https://blog.hiya.com/tag/deepfake-voice-detection) [Hiya Protect](https://blog.hiya.com/tag/hiya-protect)

# Independent benchmark: Hiya is the fastest production-ready deepfake detector

[David Martinez Gonzalez](https://blog.hiya.com/author/david-martinez-gonzalez)

Nov. 19, 2026

[![](https://blog.hiya.com/hubfs/icon-fb.svg)](https://www.facebook.com/sharer/sharer.php?u=https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector) [![](https://blog.hiya.com/hubfs/icon-x.svg)](https://twitter.com/intent/tweet?url=https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector&text=Independent+benchmark%3A+Hiya+is+the+fastest+production-ready+deepfake+detector) [![](https://blog.hiya.com/hubfs/icon-linkedin.svg)](https://www.linkedin.com/shareArticle?mini=true&url=https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector)

## At a glance

- **Independent result:** On the October 2026 Podonos Deepfake Audio Detection Benchmark, an independent public benchmark that keeps its labels private and scores every submission itself, Hiya’s deepfake detection scored 99.14% accuracy and ranked #5 of 28 systems.
- **Fastest top performer:** Hiya returned a result in 31 ms per file on average (a real-time factor of 0.011), the fastest of the five systems above 99% accuracy.
- **Low false alarms:** Hiya flagged real speech as fake 0.5% of the time, about 1 in 200 genuine clips. By comparison, all ten open-source detectors on the leaderboard scored between 47.6% and 70.5% accuracy.
- **Proven twice in 2026:** This is Hiya’s second top-tier independent result this year, after ranking #1 in the Average Result on the Hugging Face Speech Deepfake Arena in February 2026, and the detector is live in production in Hiya Protect for mobile operators.

![Independent benchmark: Hiya is the fastest production-ready deepfake detector](https://blog.hiya.com/hubfs/Podonos%20blog%20hero.png) 

It takes just a few seconds of recorded audio to clone someone’s voice, and scammers are using those clones on live phone calls to sound like a subscriber’s bank, employer, or family. According to Hiya’s [2026 State of the Call report,](https://www.hiya.com/state-of-the-call) one in four households in the six countries surveyed lost money to a phone scam last year, with rates ranging from 13% in the UK to 73% in France. Additionally, the Global Anti-Scam Alliance reported that [scammers took an estimated $442 billion](https://www.eweek.com/news/ai-scam-losses-442-billion/) from consumers worldwide in a single year and cited AI as one of the reasons making scams easier to run. 

Deepfake detection is the AI that listens to a call as it happens and decides whether the voice on the line is a real person or a synthetic one. This deepfake detection allows us to catch scammers while the call is still happening and before someone falls victim to fraud. 

### **Why independent benchmarks matter**

If you’re evaluating deepfake detection, the accuracy and speed figures you’re shown usually come from the provider’s own tests. That makes them hard to trust and hard to compare. That’s why independent benchmarks are crucial. They provide real results. Podonos is an independent third party that runs a public benchmark for audio deepfake detection. It keeps the correct labels private and scores every submission itself, so vendors cannot tune to the test or pick which numbers to share. Each system runs on the same 4,524 clips of real and synthetic speech.

In October 2026, Podonos published the results from their public benchmark study, evaluating 28 different systems. In this benchmark Hiya:

- Scored 99.14% accuracy in detecting deepfakes, one of only five systems above 99%
- Flagged real speech as fake just 0.5% of the time
- Returned an answer in 31 ms per file, the fastest of the five systems above 99% and the fastest of any system above Podonos’ 95% production bar

> *“The voice cloning tools people are using to scam each other are constantly changing. If your detector was trained once and shipped, it’s already behind. Being in production means we’re never done. We retrain on what the networks show us, and the benchmark result is proof the loop is working.”*
> 
> *Manel Terraza Farre, Director of Product, Hiya*

![leaderboard\_accuracy](https://blog.hiya.com/hs-fs/hubfs/leaderboard_accuracy.png?width=1800&height=1725&name=leaderboard_accuracy.png)

*Source: Podonos, October 2026.*

The top of the leaderboard is tight: the five systems above 99% accuracy are separated by less than one percentage point. Speed is where they pull apart, and it is where Hiya leads that group.

A real-time factor of 0.011 means Hiya analyzes a clip in about 1% of the clip's own length. A false positive rate of 0.5% means about 1 in 200 genuine clips is flagged in error.

That combination is what live calls need. A mobile operator running deepfake detection across its network with Hiya Protect needs answers fast enough to act while the call is still happening, without flagging the people it is meant to protect.

### **How Podonos tests detection**

The [Deepfake Audio Detection Benchmark](https://github.com/podonos/audio-dfd-benchmark) was built to solve a known problem: self-reported detection scores are hard to trust. Models can overfit to public test sets, and vendors can choose which results to publish. Podonos runs every system against the same fixed set:

- 4,524 audio clips across six file formats, split evenly between real and synthetic speech
- Real speech from three public corpora: VCTK, LJSpeech and LibriTTS-360
- Synthetic speech from about 25 commercial text-to-speech and voice-cloning models, including ElevenLabs, F5-TTS and Chatterbox

The benchmark reports accuracy, F1-score, false positive and false negative rates, latency and real-time factor. It does not report Equal Error Rate (EER), which assumes an ideal decision threshold that cannot be set in production. That makes it a different test from the Speech Deepfake Arena, which ranks systems by EER.

### **Built for telephony from the start**

This is the second independent evaluation this year where Hiya's detector, which is live in production today, has placed in the top tier. In February 2026, Hiya ranked #1 in the Average Result and #2 in the Pool Result on the [Speech Deepfake Arena](https://huggingface.co/spaces/Speech-Arena-2025/Speech-DF-Arena) hosted on Hugging Face, with an Average EER of 2.113% across 14 datasets. Several of those datasets include narrowband phone channels, VoIP codecs and compression artifacts. You can read the full results in our [earlier post.](https://blog.hiya.com/hiya-delivers-top-tier-speech-deepfake-detection-on-the-hugging-face-arena?hsLang=en)

Those results come from how we train the system:

- **Telephony-focused data.** We use proprietary datasets covering narrowband and wideband channels, VoIP infrastructure and a broad range of commercial codecs.
- **Quality safeguards.** In production, the system sets aside audio that is too noisy or too short to judge reliably, rather than return a low-confidence answer.

### **What the benchmark tells buyers**

The results hold three lessons for anyone evaluating detection vendors.

Accuracy alone does not tell the whole story. A false positive flags a real caller as a deepfake, which can block a legitimate customer or family member. A false negative lets a synthetic voice through, which can let fraud succeed. The right balance depends on the use case, which is why Hiya lets customers adjust detection sensitivity to their own risk tolerance.

**Speed decides whether a warning is useful.** Detection that runs slower than real time cannot flag a deepfake while the conversation is still happening. The benchmark reports latency for every system, so buyers can weigh speed alongside accuracy.

Off-the-shelf models do not keep up with modern voice cloning. All ten open-source detectors on the leaderboard scored between 47.6% and 70.5% accuracy, and most sit close to a coin flip on a balanced set. Most of these models are also built and benchmarked on clean studio audio, and their performance drops further once speech passes through narrowband telephone channels, codec compression and packet loss. Production-grade detection needs ongoing training against current voice-cloning tools, combined with deep expertise in how telephone networks shape audio.

So ask any vendor: How does your system perform on phone-quality audio? Has it been tested independently? What are its false positive and false negative rates, and how fast does it respond?

### **Proven in the open**

When evaluating a detector, a system vetted by independent parties is key and enables us to ensure we are continuously improving and adapting to stay ahead of fraud.

We are proud of the AI research, engineering and product teams behind this result, and grateful to Podonos and every research and industry group contributing to open benchmarks.

Learn more about [Hiya AI Voice Detection](https://www.hiya.com/solutions/operators), or contact us to see how it performs on your calls. 

### **Frequently Asked Questions**

What is the Podonos Deepfake Audio Detection Benchmark?  
It is a neutral, public benchmark that scores audio deepfake detectors against privately held labels. It includes 4,524 clips of real and synthetic speech, plus two telephony tracks that test the same audio over real phone codecs.

How accurate and fast is Hiya's deepfake detection?  
On the October 2026 Podonos benchmark, Hiya scored 99.14% accuracy with an F1-score of 0.991, a 0.5% false positive rate and a mean latency of 31 ms per file. It was the fastest of the five systems that exceeded 99% accuracy.

Why does telephony matter for deepfake detection?  
Most voice fraud happens over phone calls. In Hiya’s [State of the Call 2026 report](https://www.hiya.com/state-of-the-call), 31% of respondents reported receiving a deepfake call. Phone networks compress and filter audio, which can hide the signals detectors rely on. A system that performs well on clean files may not perform the same way on a live call.

How is the Podonos benchmark different from the Speech Deepfake Arena?  
The Speech Deepfake Arena aggregates 14 public datasets and ranks systems by Equal Error Rate. The Podonos benchmark uses one fixed set with private labels and reports accuracy, error rates and speed at a real operating threshold. The two are complementary, not interchangeable.

Where does Hiya use deepfake detection?  
Deepfake detection is used in both our Hiya Protect product, which operators deploy in their networks, and [Hiya Hifi](https://www.hifiapp.com/), our consumer app. Two features rely on it directly: Hiya Intelligent Call Screener, which checks if an unknown call might be a deepfake before the person decides to answer. Hiya In-Call Scam Protection, which uses it to detect whether a live call is a deepfake and alerts the call receiver before they fall victim to fraud. 

[![banner-hiya-reg](https://blog.hiya.com/hubfs/banner-hiya-reg.jpg)](https://www.hiya.com/products/connect/number-registration)

### Latest Articles ⚡️

- <https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector>
  
   Independent benchmark: Hiya is the fastest production-ready deepfake ...
  
  David Martinez Gonzalez
  
  Nov. 19, 2026
- <https://blog.hiya.com/leveling-up-uk-scam-defense-how-o2-made-the-move-from-flagging-to-outright-blocking?hsLang=en>
  
   Leveling up UK scam defense: how O2 made the move from flagging to ...
  
  Camilla Dickson
  
  Nov. 9, 2026
- <https://blog.hiya.com/which-european-country-has-the-worst-phone-spam-problem?hsLang=en>
  
   Which European country has the worst phone spam problem?
  
  Jonathan Nelson
  
  Nov. 4, 2026

## Related articles

<https://blog.hiya.com/leveling-up-uk-scam-defense-how-o2-made-the-move-from-flagging-to-outright-blocking?hsLang=en> ![](https://blog.hiya.com/hubfs/Operators-Hiya_VMO2-UK.webp)

## [Leveling up UK scam defense: how O2 made the move from flagging to ...](https://blog.hiya.com/leveling-up-uk-scam-defense-how-o2-made-the-move-from-flagging-to-outright-blocking?hsLang=en)

Camilla Dickson

Nov. 9, 2026

<https://blog.hiya.com/which-european-country-has-the-worst-phone-spam-problem?hsLang=en> ![](https://blog.hiya.com/hubfs/Blog-Which-European-country-has-the-worst-phone-spam-problem-b.webp)

## [Which European country has the worst phone spam problem?](https://blog.hiya.com/which-european-country-has-the-worst-phone-spam-problem?hsLang=en)

Jonathan Nelson

Nov. 4, 2026

<https://blog.hiya.com/telefonica-launches-with-hiya-in-spain?hsLang=en> ![](https://blog.hiya.com/hubfs/Operators-Linkedin%20social.webp)

## [Solving the unknown number problem in Spain with Telefónica](https://blog.hiya.com/telefonica-launches-with-hiya-in-spain?hsLang=en)

Camilla Dickson

Oct. 20, 2026

[![](https://blog.hiya.com/hubfs/hiya-new-purple.svg)](https://www.hiya.com/)

[![](https://blog.hiya.com/hubfs/linkedin-1.svg)](https://www.linkedin.com/company/hiyainc) [![](https://blog.hiya.com/hubfs/facebook-1.svg)](https://www.facebook.com/hiyainc/)

###### Products

- Connect
- [Number Registration](https://www.hiya.com/products/connect/number-registration)
- [Branded Call](https://www.hiya.com/products/connect/branded-call)
- [View Plans](https://www.hiya.com/products/connect/pricing)
- Protect
- [Spam Analytics](https://www.hiya.com/products/protect/spam-analytics)
- [AI Voice Detection](https://www.hiya.com/products/protect/ai-voice-detection)
- Apps
- [Hiya Spam Blocker](https://www.hiya.com/products/apps/hiya-spam-blocker)
- [Hiya AI Phone](https://www.hiya.com/products/apps/hiya-ai-phone)

###### Solutions

- Company size
- [Enterprise](https://www.hiya.com/solutions/enterprise)
- [Call Centers](https://www.hiya.com/solutions/call-center)
- [Small and Medium](https://www.hiya.com/solutions/businesses)
- Service providers
- [Operators](https://www.hiya.com/solutions/operators)
- [OEMs and technology](https://www.hiya.com/solutions/technology-partners)

###### Considering Hiya?

- [Why Hiya](https://www.hiya.com/why-hiya)
- [How it Works](https://www.hiya.com/how-it-works)
- [Customer Stories](https://www.hiya.com/case-studies)
- [Voice Intelligence Platform](https://www.hiya.com/why-hiya/voice-intelligence-platform)
- [Trust Center](https://www.hiya.com/why-hiya/trust-center)
- [Modern Slavery](https://www.hiya.com/company/modern-slavery)
- [About Hiya](https://www.hiya.com/company/about)
- [Careers: We're hiring!](https://www.hiya.com/company/careers)
- [Contact us](https://www.hiya.com/contact-us)

###### Resources

- [Resource Center](https://www.hiya.com/resources)
- [Partner Program](https://partners.hiya.com/)
- [Get Support](https://hiya.com/support)
- [Developer Docs](https://developer.hiya.com/)
- [Hiya Blog](https://hiya.com/blog)
- [Events](https://www.hiya.com/events)
- [Press Kit](https://www.hiya.com/newsroom#press-kit)

![ANAB](https://blog.hiya.com/hubfs/logo-anab.svg) ![ISO 27001](https://blog.hiya.com/hubfs/logo-iso27017.png) ![AICPA](https://blog.hiya.com/hubfs/logo-aicpa.svg) ![IAF](https://blog.hiya.com/hubfs/logo-iaf.svg)

[Privacy Policy](https://www.hiya.com/legal/privacy) [Terms of Service](https://www.hiya.com/legal/terms-of-service) [App Data Privacy Policy](https://www.hiya.com/legal/app-data-protection-and-privacy-policy)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "David Martinez Gonzalez",
    "url" : "https://blog.hiya.com/author/david-martinez-gonzalez"
  },
  "dateModified" : "2026-11-19T20:32:06.276Z",
  "datePublished" : "2026-11-19T20:32:06.000Z",
  "headline" : "Independent benchmark: Hiya is the fastest production-ready deepfake detector",
  "image" : [ "https://blog.hiya.com/hubfs/Podonos%20blog%20hero.png" ],
  "mainEntityOfPage" : {
    "@id" : "https://blog.hiya.com/independent-benchmark-hiya-is-the-fastest-production-ready-deepfake-detector",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://blog.hiya.com/hubfs/hiya-logo-3.svg"
    },
    "name" : "Hiya Inc."
  }
}
```