> For the complete documentation index, see [llms.txt](https://bellingcat.gitbook.io/toolkit/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://bellingcat.gitbook.io/toolkit/more/all-tools/hiya-deepfake-voice-detector.md). # Hiya Deepfake Voice Detector A tool for detecting audio deepfakes. ## URL ## Description Hiya identifies deepfake audio. A deepfake is [synthetic](https://www.turing.ac.uk/blog/what-are-deepfakes-and-how-can-we-detect-them) media, such as audio, video, or images, that can mimic the appearance or voice of a real person by using artificial intelligence. This technology has been misused to spread disinformation and scam people, for example, by faking audio messages from [officials](https://www.abc.net.au/news/2024-01-23/ai-phonecall-posing-as-joe-biden-being-investigated/103378702) or [loved ones](https://archive.fo/cn2UV). Hiya is available as both a standalone [Chrome](https://chromewebstore.google.com/detail/hiya-deepfake-voice-detec/akmieeldmgcllmokbpaibfelofjiilpc) extension and an integrated tool within the [InVID Verification Plugin](https://bellingcat.gitbook.io/toolkit/more/all-tools/invid), a suite of verification tools used by journalists, researchers and fact-checkers to verify digital media. Once installed in the browser, the extension aims to detect audio deepfakes across social media, news, and other websites. To analyze audio in a video on any website, simply open that page and click the extension icon in the Chrome browser in the top right corner. Play the video, then click the “start analyzing” tab to collect the audio sample for analysis. The media being analyzed and the extension work side by side without the need to close either.

Click on the "start analyzing" option to collect and analyze the media of your choice.

The tool then collects and rates the audio on a scale of 1 to 100 points. The lower the score, the higher the likelihood that the sample is fake. The closer the score is to 100, the higher the probability that it is real. For instance, the voice of Galen Reich, an investigative technologist at Bellingcat, in a YouTube video "[How To... Track Your Satellite Searches](https://www.youtube.com/watch?v=cKNltv2PExo)" from Nov. 28, 2024, analyzed via Hiya has been rated authentic.

Hiya collects an audio sample from the YouTube clip while it is being played.

Hiya rates the sample after collection, which in this case is an authentic audio.

Our testing shows the extension works on almost any website, including social media sites like X (formerly Twitter), Instagram, TikTok, Facebook and Bluesky, provided the websites are opened in Chrome and the tool is installed.

Hiya's Chrome extension shows a score of 3/100 ("likely a deepfake") for a portion of a video (from 0:17 to0:25), featuring the Indian Army Chief, purportedly admitting that India lost six jets to Pakistan in the India-Pakistan conflict. The video was circulating online in August 2025. The author of this text investigated this synthetic claim by using - amongst other methods - Hiya's Chrome extension. Other organizations, for example CyberPeace and Boom, had also fact-checked the video.

The Chrome extension needs to check just [one second](https://blog.hiya.com/hiyas-new-chrome-extension-identifies-deepfakes) of audio or video to determine whether the tool considers it as fake or real, while the version integrated into the InVID Verification Plugin allows longer media uploads (audio files only) ranging from two seconds to five minutes.

Hiya is integrated within the InVid Verification Plugin that supports media uploads ranging from two seconds to five minutes.

The extension uses "[Hiya’s AI Voice detection capabilities](https://chromewebstore.google.com/detail/hiya-deepfake-voice-detec/akmieeldmgcllmokbpaibfelofjiilpc)" which are "advanced AI solutions powered by deep learning," [according](https://www.hiya.com/products/protect/ai-voice-detection) to the tool provider. It further [states](https://chromewebstore.google.com/detail/hiya-deepfake-voice-detec/akmieeldmgcllmokbpaibfelofjiilpc?pli=1): "Please note that this extension utilizes probabilistic algorithms, which means that its predictions or outputs are based on statistical analysis and may not always be 100% accurate." We were not able to find more detailed public information about the inner workings of Hiya. The tool provider also states its deepfake detection [technology ](https://www.hiya.com/newsroom/press-releases/hiya-acquires-deepfake-voice-detection-leader-loccus-ai)is able to analyze audios and videos in multiple languages. It is important to note that the tool is not perfect and may lead to false positives. Hiya can nevertheless be a handy tool when researching the rising tide of AI generated disinformation. Several fact-checking teams, including [AFP Fact Check](https://factcheck.afp.com/doc.afp.com.78448G3), can be seen using Hiya, especially the version integrated within InVid, as part of their research workflows. ### Cost * [x] Free * [ ] Partially Free * [ ] Paid After signing up for the extension, users can do 200 daily checks for free. ## Level of difficulty


1

## Requirements A sign up via email is required to use Hiya's Chrome extension. ## Limitations 1. As deepfake technology advances, it is becoming [increasingly](https://www.thebureauinvestigates.com/stories/2026-07-30/deepfake-experts-whats-real-sam-gregory) difficult for tools to detect fake media. Hiya may not always detect deepfake audio or video. False positives or negatives are likely to occur. 2. The Chrome extension just analyses small parts of any video or audio and may therefore not be the best choice for analyzing longer audios and videos. Switching to the Hiya version within the [InVid](https://bellingcat.gitbook.io/toolkit/more/all-tools/invid) plugin could be one option to mitigate this since it allows larger file uploads. However, even the InVid version suggests to "Test smaller sections for a more precise analysis.” 3. The tool only provides a probability score of how likely it is that the sample is fake or real but it does not provide any contextual information. It is crucial to find other ways of verification, such as reaching out to experts, finding contextual information in news reports, or using other deepfake detection tools to double-check Hiya's results. ## Ethical Considerations The tool may produce false positives or negatives. Users, especially journalists and researchers, need to be aware that it is crucial to double-check any results before drawing conclusions. It is recommended to be transparent about the limitations of the tool when including Hiya scores in publications or when communicating them to others. The tool provider itself states that since the tool uses "[probabilistic algorithms](https://chromewebstore.google.com/detail/hiya-deepfake-voice-detec/akmieeldmgcllmokbpaibfelofjiilpc?pli=1)", the output is based on statistical analysis, so the results may not always be accurate. Therefore it is important to "always verify important information independently." Access to detection tools may help malicious actors improve evasion [strategies](https://arxiv.org/pdf/2102.06109), increasing the sophistication of future deepfakes. Testing shows the tool is able to analyze audio from private chats from platforms like WhatsApp and Discord if opened in Chrome. This access raises privacy concerns and thus be extra careful about using the tool on chats as sensitive information may accidentally end up being collected and analyzed by the tool. ## Guides and articles Henk Van Ess wrote the guide, “[Reporter’s Guide to Detecting AI-Generated Content,](https://gijn.org/resource/guide-detecting-ai-generated-content/)” for the Global Investigative Journalism Network (GIJN) on Sept. 1, 2025. The guide recommends using Hiya’s Chrome extension, alongside techniques for detecting and debunking deepfakes. Also check out DW’s article “[Synthetic Audio Detectors Put to the Test](https://innovation.dw.com/articles/synthetic-audio-detectors-tested)” from Sept. 29, 2025, which tested Hiya, integrated within InVid, alongside deepfake detection tools Deepfake-O-Meter and Deepfake Total. ## Tool provider Hiya, [Seattle](https://chromewebstore.google.com/detail/hiya-deepfake-voice-detec/akmieeldmgcllmokbpaibfelofjiilpc), U.S. ## Similar tools A tool similar to Hiya is [Deepfake-O-Meter](https://bellingcat.gitbook.io/toolkit/more/all-tools/deepfake-o-meter), which is an initiative of the [University at Buffalo](https://www.buffalo.edu/home/story-repository.host.html/content/shared/university/news/ub-reporter-articles/stories/2024/09/deep-fake-o-meter.detail.html). While Hiya only verifies deepfake audio or AI-generated voice in a video, Deepfake-o-Meter allows users to analyze deepfake audio, video, and images. Both can be used simultaneously to cross-check the same sample, thus reducing the risk of false positives or false negatives. Hiya and DeepFake-O-Meter are free to use, but they require users to sign up. Hiya’s Chrome extension verifies content across social platforms, websites, etc., allowing users to check for deepfake voices or AI-generated media directly in their browser. Whereas DeepFake-O-Meter requires users to download media and upload it manually to the platform for analysis. Meanwhile, the InVID plugin that integrates Hiya offers additional flexibility, allowing users to either paste media links or upload files directly from their computers, but a sign up is required to use it. Another tool for audio deepfake detection and can be combined with others for effective analysis is [Deepfake Total](https://deepfake-total.com/) by Cognitive Security Technologies (CST) department of Fraunhofer AISEC ([Fraunhofer Institute for Applied and Integrated Security](https://www.aisec.fraunhofer.de/)), a research institute from [Germany](https://www.aisec.fraunhofer.de/de/impressum.html). This tools [supports](https://deepfake-total.com/) analyzing links from Instagram, YouTube, X or local file uploads. | Page maintainer | | --------------- | | Anisa Shabir. | | October 2026 | --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://bellingcat.gitbook.io/toolkit/more/all-tools/hiya-deepfake-voice-detector.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.