Lightspeed Ventures-backed audio platform Pocket FM introduced it has partnered with voice-cloning firm ElevenLabs to rapidly convert textual content content material, equivalent to script, into audio collection utilizing AI.
Pocket FM, which raised $103 million in Collection D funding in March, advised TechCrunch on the time that it was already experimenting with the flexibility to transform textual content content material into audio utilizing ElevenLabs‘ tech. Now, the India-based firm has expanded the partnership to make the conversion instrument obtainable to all creators over the subsequent few weeks.
Within the check part, Pocket FM already produced 30,000 hours of audio collection utilizing ElevenLab’s AI tech. With the brand new roll-out, the startup expects to triple its content material library of over 100,000 hours of audio content material this 12 months. Pocket FM additionally mentioned that through the experimental part, the AI-powered instruments helped it lower the price of producing audio by 90%.
Pocket FM’s co-founder and CTO Prateek Dixit advised TechCrunch over a name that with this partnership, the corporate needs to make it simpler for writers to transform their writings into audio collection.
“We have over 250,000 writers (including the ones on the company’s Pocket Novel writing plaform) and this partnership decreases the cost of setting up and recording audio for them,” he mentioned.
“Even with a good set up of recording tools and equipment, writers can produce roughly 30 minutes of high-quality audio content per day. With the AI tools, this output can be 10 times more,” he added.
Pocket FM has constructed a instrument integrating ElevenLabs tech, via which it’s providing 50 voices for writers who wish to convert their content material. ElevenLabs’ co-founder Mati Staniszewski mentioned that his firm’s instrument understands the context of the writing and infers feelings via the voice robotically.
“Working with Pocket FM, we are deploying our newer models that understand the genre of writing and are emotionality better,” Staniszewski mentioned.
Dixit famous that based mostly on information from customers’ engagement with this type of content material, the platform additionally plans to counsel voices that work effectively for writers in a specific style.
Pocket FM will not be the one audio collection platform experimenting with AI-powered instruments. Google-backed Kuku FM is utilizing GPT-4, Claude, BandLab and even ElevenLabs to assist its writers with completely different phases of creation, together with refining script, producing thumbnails, including sound results and changing textual content into audio.
Kuku FM advised TechCrunch that it is usually experimenting with utilizing visible era instruments equivalent to Midjourney and Runway to create adverts associated to content material.
High quality of content material and impression on artists
The promise of AI-powered instruments is to generate extra content material sooner, however that doesn’t imply the content material is nice. Pocket FM’s reply to aiding discovery and surfacing high quality content material is making its discovery algorithm subtle and experimenting with consumer engagement.
“If a writer publishes an audio series, we surface that content to a select number of users and observe engagement metrics. If these metrics are positive, we further propagate that,” Dixit mentioned.
Using AI might result in faster outcomes and a much bigger content material library for these platforms, however it’s going to additionally cut back the roles of voiceover artists working with them. India’s Affiliation of Voiceover Artists (AVA) has expressed its issues about AI taking on.
“If AI takes over, we are finished. As voice artists, we need to get some regulation in place so that our livelihood is protected,” Amarinder Singh Sodhi, the affiliation’s basic secretary, advised Indian publication Scroll.
Sodi additionally advised Scroll about incidents the place voiceover artists have been referred to as into the studio to document samples to coach AI with out acquiring their consent or informing them.
“On an emotional level, it scares me. By using AI, you are essentially diluting the human experience of storytelling. You lose out on an emotional connection,” Delhi-based voiceover artist Aditya Mattoo advised TechCrunch.
He added that giving entry to premium voices to individuals who don’t have the style and talent to provide high quality content material will result in the market getting flooded by unhealthy content material.
Once we requested in regards to the impression of AI-powered voice era on Pocket FM, the corporate didn’t instantly reply the query. Nonetheless, Dixit famous that engagement with AI-generated content material in its experiments is “as good as human voiceover production.” Notably, the corporate can also be engaged on expertise to include a number of voices in a single audio output.
Each Pocket FM and Kuku FM don’t at the moment label their content material to point if AI has been used within the creation course of.
