Skip to main content

Audio Intelligence Changelog

Message Verification API Removal

🗑️ Deprecations​

  • Message Verification API removal on 10 November 2026: The Message Verification API will be removed on 10 November 2026. After that date, the following will no longer be available:

    • The message verification endpoints under /spaces/{owner}/{space}/verifications/message, including list, create, get and update.

    Requests to the removed endpoints will return a not-found error. The phone and other verification APIs (identity, authenticity, voiceprint and their related endpoints) are not affected.

Audio Intelligence Digital Models Retired

🤖 Models​

  • Digital models retired: The digital voiceprint, identity, authenticity and message verification models are no longer available. Requests that select digital model return a model-not-available error. Use the phone models instead, which support audio with sample rates of 8 kHz or higher. For their specifications, please visit our model index page.

Audio Intelligence 2026.04.r1p0

🤖 Models​

  • Introducing Phone/v8: We are proud to announce our newest model, Phone/v8. This model delivers improved accuracy and broader coverage of newer deepfake attack techniques compared to its predecessor. It also includes an updated voice activity detector, improving voice detection accuracy for phone-channel audio. For comprehensive technical specifications and benchmarks, please visit our model index page.

Audio Intelligence 2026.03.r1p0

🤖 Models​

  • Introducing Phone/v7: We are proud to announce our newest model, Phone/v7. This model delivers improved accuracy and broader coverage of newer deepfake attack techniques compared to its predecessor. For comprehensive technical specifications and benchmarks, please visit our model index page.

Audio Intelligence 2026.02.r1p0

🤖 Models​

  • Introducing Phone/v6: We are proud to announce our newest model, Phone/v6. This model delivers improved accuracy and broader coverage of newer deepfake attack techniques compared to its predecessor. For comprehensive technical specifications and benchmarks, please visit our model index page.

💥 Changes​

  • Removal of Phone/v5: The Phone/v5 model has been replaced by Phone/v6 and is no longer available on the platform. Please update your integrations to use Phone/v6.

Audio Intelligence 2025.11.r1p0

✨ Features​

  • Next-Generation Phone Authenticity Verification: Introducing phone/v5, the new default phone model for authenticity verification. This advanced model delivers enhanced accuracy with double the parameters compared to previous versions, providing superior deepfake detection capabilities for telephone channel audio.

🛠️ Improvements​

  • Enhanced Detection Performance: The phone/v5 authenticity verification model has been optimized with an expanded parameter set, enabling more sophisticated analysis of voice authenticity and improved reliability in identifying AI-generated or modified audio over telephone channels.

Audio Intelligence 2025.08.r1

🤖 Models​

  • Introducing Digital/v3 authenticity verification model: We are proud to announce our newest model, Digital/v3. This new model deliver improved accuracy, with coverage of additional attacks and newer techniques. For comprehensive technical specifications and benchmarks, please visit our model index page.

Audio Intelligence 2025.04.r1p0

🤖 Models​

  • Introducing Phone/v3: We are proud to announce our newest model, Phone/v3. This new model deliver enhanced performance with improved accuracy compared to their predecessor. For comprehensive technical specifications and benchmarks, please visit our model index page.

✨ Features​

  • Streaming Demo Experience: Our web platform now offers a new streaming mode, in which the user can experience how the verification results are delivered in real time.
  • Comprehensive Verification Mode: The platform console now displays the mode in which verifications are performed, either via the API or the web demo.

Audio Intelligence 2025.03.r1p0

✨ Features​

  • Enhanced Demo Experience: Our web platform now offers a streamlined user experience with the ability to select between Voice Authentication and AI Voice Detection demonstration workflows, allowing for more targeted exploration of our capabilities.
  • Comprehensive Identity Analytics: The platform console has been enhanced to display a complete set of identity verification metrics, including detailed replay and synthesis scores, providing deeper insights into verification results.

🛠️ Improvements​

  • Refined Identity Verification Models: The phone/v1.1 and digital/v1.1 identity verification models have been optimized to perform synthesis and replay detection with improved accuracy and reliability, ensuring more robust security assessments.