THD Podcast 141 - HEAD Acoustics: What Happens When Audio Systems Outsmart Your Test Signals?

The Total Harmonic Discussion / the THD Podcast. A weekly discussion on audio and headphone technologies and the people who bring the technology to market.In episode 141 of the THD Podcast, episode 141 of the THD Podcast, Jacob Soendergaard joins from HEAD acoustics GmbH to talk about ETSI's new Speech-Based Distortion Measure may be the industry's answer to confused Automated Speech Recognition (ASR's).Standardized as ETSI TS 104 063. As audio systems become increasingly optimized to handle speech and powered by AI, traditional distortion metrics are reaching their limits. This new framework is the first metric designed to quantify distortion using natural human voice signals instead of artificial tones. It came about because modern smart speakers, ANC headphones, and smartphones rely on complex, non-linear AI processing and noise-cancellation algorithms. Generic sine waves just don't produce a reliable test to learn what happens and whether the ASR will hallucinate as to what you are saying. An interesting and timely discussion so we hope you enjoy it. Check out the podcast on YouTube  Spotify⁩  Amazon Music Rumble  iHeartRadio⁩ Apple Google Thanks to Barry Vogel and the ALTI Audio & Loudspeaker Technologies International for their supportThanks to audioXpress for their support on this episode as well! #audioengineering #acoustics #ETSITS104063 #etsi #HEADacoustics #ETSISTQ #NonlinearityMeasure #SpeechQuality #voicequality #AudioTesting #AcousticTesting #SignalProcessing #Electroacoustics #HATS #ACQUA #LabCore Please like, subscribe, share, and comment with your feedback! Enjoy the podcast!Support the podcast on our Patreon page: www.patreon.com/THDPodcastwww.db-ent.comlinkedin.com/in/DavidJLindberg @dlindberg11 discord.gg/HF6NF3fchttps://www.head-acoustics.com/www.totalharmonicdiscussion.com0:00 Introduction to audio testing0:24 Podcast opening and context2:13 The problem with pure tones5:46 Introducing nonlinear measurement8:10 The mechanics of NLM analysis11:00 Standardization and metrics12:40 Interpreting NLM output15:11 System-level testing nuances18:51 Handling delay and complexity22:40 Computational efficiency and summary