Introducing Phonexia Voice Inspector 6: Ready for Deepfakes
Forensic audio analysis keeps evolving, and so does Voice Inspector. With this new release, we're introducing our biggest update yet, headlined by Deepfake Detection technology. From now on, your forensic team will be ready for the rise of AI voice cloning and cases where authenticity is in question. Here's what's new.
Deepfake Detection: A New Line of Forensic Analysis
Voice Inspector 6 introduces our Deepfake Detection technology, built to help identify whether an audio sample may contain artificially generated speech. It's designed for cases where the authenticity of a recording itself is in question.
The tool evaluates audio using a Log-Likelihood Ratio (LLR) score, a statistical measure of how strongly the evidence points toward one of two possibilities: that the audio is synthetic, or that it's a genuine human recording.
Positive scores point toward synthetic or manipulated audio. The higher the value (typically up to around +10), the stronger the evidence of a deepfake.
Negative scores point toward authentic speech. The lower the value (typically down to around -10), the stronger the evidence the recording is genuine.

Deepfake detection results are displayed as a Deepfake Score under the Deepfake results tab.
We also want to set expectations from the start: deepfake technology won't stand still, but neither will we. We're committed to the continuous development of Deepfake Detection, releasing regular updates to keep pace with new synthesis techniques as they emerge, and sharpening the explainability behind every result we deliver.
Join Free Live Webinar and See Voice Inspector 6 in Action
We'll walk you through all the new features on our free webinar on the 23rd of September at 2:00 PM CEST. Register below to see the new Voice Inspector in action and get answers to any of your questions.
Many-to-Many (M:N) Voice Comparison
Previous versions of Voice Inspector supported:
1:1 voice comparisons (one suspect against one questioned recording)
and 1:N voice comparisons (one suspect against multiple questioned recordings).
But when investigators needed to check multiple suspects against multiple recordings, they had to run each voice comparison separately and piece the results together by hand.
Voice Inspector 6 solves this with M:N voice comparison. Now you can compare any number of suspects against any number of questioned recordings in a single analysis, generating one summary with results for every suspect-recording pair.
Example: Investigator working on an organized crime case with three intercepted calls and four suspects can now run all combinations at once instead of four separate analyses.

You’ll find the results of all possible combinations of suspects and recordings in a clear table.
Voice Comparison Without Population Sets
Voice Inspector's standard comparison method, built around ENFSI guidelines and population sets, remains the recommended approach and isn't going anywhere. But we know that a suitable population set isn't always available, and sometimes a quick preliminary read is all that's needed.
That's why 6 introduces an additional comparison mode that works without population data. It produces an Evidence score (also on a log-likelihood-ratio scale) reflecting the similarity between a questioned recording and a suspect, calibrated on broad, general population data.
This option is meant to complement the ENFSI-compliant workflow, not replace it. It gives you more flexibility when circumstances call for it.

When using voice comparison without population sets, you’ll always see the blue bar on the left as a reminder.
Fresh Interface & Modernized Foundation
Voice Inspector 6 comes with a modernized interface, which includes new icons and a cleaner, more intuitive design. It's still the same application, just easier on the eyes.
We've also modernized the technical foundation of the application. These changes won't be visible day-to-day, but they set Voice Inspector up for a more sustainable future and let us ship updates more frequently going forward.
Join Free Live Webinar and See Voice Inspector 6 in Action
These new Voice Inspector features are big. Join us for an exclusive free live webinar on the 23rd of September at 2:00 PM CEST, where our forensic experts will:
show you the Deepfake Detection tool in action,
demonstrate M:N analysis live,
answer your technical and licensing questions.
We Want to Hear From You
This release marks a fresh chapter for Voice Inspector, and we plan to keep the momentum going as deepfakes evolve. If you have feedback on this version or any ideas for what you'd like to see next, we'd love to hear from you.
To upgrade to Phonexia Voice Inspector 6, please register for the webinar above or reach out to your sales contact.
If you're not using Phonexia Voice Inspector yet and would like to try it without obligations, feel free to schedule a live demo.


