Lovo launches AI ‘voice-over’ platform that creates realistic human voices
Lovo, Inc., an AI voice-over platform developed by a team of AI and machine experts from the Univerity of California-Berkeley, has launched what the company describes as “a human-like voice-over platform” designed for education, marketing, entertainment, and other audio content.
The company is presently targeting its Lovo Studio platform for businesses, governments, and other entities, “given the current need for ‘distance narration’ during this period of social distancing” in response to the Covid-19 pandemic.
“LOVO Studio provides many options for getting voice-over projects done during these tough times of working remotely, creating marketing videos, educational materials, corporate videos, and other voice-over projects,” which have “been very difficult” with so many governments and business offices shuttered, the company emphasized in its announcement.
The company said its new Lovo Studio is a sophisticated and easy-to-use platform using AI to … recreate a human voice with emotion and tone gradations which make either cloned or synthesized voices sound very realistic.”
With the “emotional range and realistic vocal characteristics” that are available, the company said Lovo Studio’s “cloned voices are practically indistinguishable from the original voice.” The “platform also provides more than 50 other voices, both computer-created and human, from which to choose for any voice work without needing a studio or expensive equipment.”
A recent graduate from the Fall 2019 UC Berkeley SkyDeck accelerator group, Lovo Studio, generates “a realistic-sounding voice ‘clone’ with only five minutes of a target voice clip,” which “during a time of social distancing … makes it very fast and easy to generate online learning materials or voice-overs for remote production projects.”
“You can hide your emotions behind words, but you can’t hide it in your voice. Human voice delivers what words or robotic sounds can’t convey,” said Tom Lee, Lovo co-founder. “We have created a one-stop-shop for audio and voice work.”
“Imagine a combination of Canva and Photoshop but for audio,” Lee said in a statement. “On one hand, you can leverage a catalog of existing voices and background music to get started with your project right away, or create your custom voice and add music of your choice to make something more personal. And then, features like pronunciation editing, emphasis, speed, and pause control allow for fine tuning your finished content.”
“On the other hand,” Lee said, you can also “easily touch-up where you need to” by filling gaps, smoothing out rough patches, redos, etc.
“For instance,” Lee explained, “if you used a professional actor for your narration and that video needs to be corrected, you won’t need to call that actor back.”
Lee added that “If you are doing a voice-over for one project, you can’t be doing narration for something else at the same time. But with LOVO, an actor could have his voice reproduced and used in multiple projects. It is practically indistinguishable from the ‘real’ voice,” and “this makes the human voice truly scalable.”
The company said its AI design team “originally started by analyzing the emotional status of customer service agents and the customers who called in with an issue, giving the engineers unique insights into how emotions affected voices.” Then, with “this understanding of the relationship between emotion and voice,” they were able to synthesize and make “computer-generated voices like Siri and Alexa sound more natural and human-like.”
The Lovo Studio team “came together after analyzing the emotional status of customer service agents and customers who called in,” the company said, so our “engineers had unique insights into how human voices reflect [a person’s] emotional state.”
Lovo Studio is an SaaS program the company said offers “flexible options and an API for easy integration,” and is currently being used “by schools, marketing agencies, customer service and HR teams, authors, software developers, and film production companies.”
Lovo said it’s currently capable of providing more than 50 synthesized and human voices to choose from on the Lovo Studio on the platform with multiple options for the voices, music, pauses, speeds, and pronunciation.
Side-by-side demos can be viewed here, and a studio promo is available here.
The system works like this:
● Sign up for the platform and choose a voice from the library of more than 50 options;
● Upload a script and a voice file is created within “seconds,” the company said;
● Fine-tune the file by adding specific features like pauses, emphasis, speed, pronunciation, pitch, and then add background music;
● Save, load, and share the project;
● Clone a voice if desired by providing a 5-minute clip.
If the company is concerned about its technology being used to generate deepfakes, it does not seem to provide any indication to that effect in its marketing materials.