Issue link: https://digital.copcomm.com/i/1521235
46 M PS E . O R G The app can also record a new performance to use in DeepEditor to train and recreate the performance from the original take. This avoids having to bring an actor in for ADR or a reshoot. In closing, Seth pointed out that most of the presentations were focused on enhancing the creative process, but at less cost. In the human role, they hope to make it simpler. Whether it's just text prompting or more graphical, they feel that professionals with years of experience will adapt to these tools quickly. Just get educated. Generative AI in Sound Moderator: Bob Bronow, Re-recording Mixer Panel Mitch Singer, Strategic Advisor, ReSpeecher Dustin Blank, Eleven Labs - Head of Partnerships Brian Riordan, Founder and CEO, Levels Audio In opening, Bob Bronow described how we searched for sound in the past (and currently). In his example, a music search, you could search by genre, instrumentation, tempo, style, etc. You'll get a list to look and listen to. With a generative AI search, you might describe the mood and emotion of the scene and the genre, instrumentation, and tempo and you'll get a bespoke piece of music which is created specifically for your scene. With Sound Effects, a similar search, but now they are working on an AI creation that will generate sound from text prompts. But not to worry yet and the program when asked for Whistling Wind, it created wind with a man whistling in the background. In Dialogue and ADR, we have new tools to separate backgrounds from voice. Also, there are tools now to replace frequencies that were not recorded. These tools will only get better. First up was Mitch Singer, Strategic Advisor of ReSpeecher. ReSpeecher is voice-cloning software. Mitch wanted to describe some terms that ReSpeecher uses. First was "Source." The "Source" is the voiceover actor that is performing the new line. Next was "Target." The "Target" is the voice to be emulated or cloned. ReSpeecher has been used on 170-plus projects, not the least is Star Wars projects. The company is also working with medical applications to give patients that lose their voice box the ability to speak with a digital version of their original voice. In addition, they are working in tandem with cyber security firms to keep cloned voices from breaching voice verification systems. Mitch then started to describe efforts they are employing to ensure that their cloning process is not abused. The first is a digital signature on every sound clip that ReSpeecher creates. This "watermark" can be used to identify the clip as artificial. They envision a future when the clips could be automatically scanned and rejected if they are not authorized. ReSpeecher is working with Partnership AI, Digital Entertainment Group, and Open Voice to create guidelines for the ethical use of voice cloning. Next Dustin Blank, Head of Partnerships for Eleven Labs, played a video of their newest product roll-out. It is a way to train your voice into Eleven Labs to create a voice library for clients to browse and purchase voices for their project. It would allow voice actors to get paid without any additional work. At its core, Eleven Labs is voice cloning AI software which has Text to Speech and Speech to Speech input. The last panelist was Brian Riordan, Founder and CEO of Levels Audio. Levels Audio is a post-production house in the middle of Hollywood focusing on documentary and unscripted projects. Brian described how his editors are using AI tools, Eleven Labs for one, to enhance and speed up their editing process. On the projects that Levels Audio is working on, Brian doesn't see these tools as replacing any of his editors. He sees the AI tools as a way for them to create better tracks for his mixers to mix. A discussion about using AI tools ensued. One example was a production that halted after shooting ended because of COVID. It resumed two years later and the 10-year-old boy, who was now 12, didn't sound like he did two years before. So, they cloned his younger voice to replace his older voice ADR with Speech to Speech with "…you're in the water on a board and the wave is coming. Are you going to paddle and go with the wave, or you going to just get crushed by it? It's coming." –BRIAN RIORDAN