AudioStack: case study
Using AudioStack’s fully-automated production process, anybody can create beautiful sounding audio in seconds. Starting with simple text, users can compose professional-quality pieces including music and complex audio, without any previous experience. AudioStack’s technology reduces the time for creating audio from days to just seconds.
Timo Kunz is the company co-founder and CEO. A musician all his life, with a background in AI and a PhD in machine learning, Timo was searching for a way to combine his passion and expertise into one business idea.
Artificial intelligence is now a major force in the communications industry, but as recently as five years ago, synthetic voices did not sound natural at all. Optimistic that the technology would improve, Timo continued to test his ideas.
I knew that there had to be this magic moment when voices would sound so good, you could finally produce nice things to listen to. Then in 2018, AI architecture made a huge step change and, all of a sudden, voices sounded beautiful and very lifelike. So I quit my job and started AudioStack.Timo Kunz Co-founder and CEO of AudioStack
From text to beautiful audio in just a few moments
AudioStack (previously Aflorithmic) offers Audio-as-a-Service for developers and content creators. It works by converting simple text into voice content, after which users can clone their own voice to use with the audio, pick a voice from the AudioStack library or choose one their customers are familiar with.
Users have complete control over the audio development workflow, from voice production to sound engineering, music and most importantly mastering. Content creators with no previous experience can compose professional-quality pieces to develop their unique audio branding, then deliver these at scale through websites, mobile apps, IoT devices or social media platforms.
“The voice is just one ingredient,” Kunz explained. “What you get from us is a full audio production, much closer to a professionally produced advert, podcast or any audio experience you can think of.”
As its AI business accelerates, AudioStack turns to Machine Intelligence Garage for support.
Initially, AudioStack raised funds through a friends and family round, before winning a grant from Innovate UK. With a funded project underway, the business took an overwhelming leap forward, so it was time to seek specialist advice.
In 2020, AudioStack joined Machine Intelligence Garage, Digital Catapult’s flagship AI accelerator programme. Machine Intelligence Garage has supported over 100 startups, helping to overcome the challenges that artificial intelligence (AI) and machine learning (ML) innovators face.
“We applied to Digital Catapult for several reasons,” said Kunz. One of them was the whole founder-to-founder support. We were not first-time founders and were experienced technologists, but in the tech startup community we were all newbies, so it was very important for us to connect to AI founders, all at the same stage and struggling with the same problems.”
The Machine Intelligence Garage programme provides AI startups with the technical support, workshops and the computation power to advance their AI solutions, including cloud credits for Amazon Web Services (AWS), which AudioStack used to accelerate testing.
“As a company we were very hungry for computing power,” said Kunz. “The credits solved a big problem for us because, at the time, our R&D team really couldn’t test enough. Suddenly, with all these credits, we quadrupled our testing speed.”
Using Digital Catapult’s Ethics Framework, Machine Intelligence Garage provides a range of tools to help startups use data and run operations more ethically and responsibly.
As a result, AudioStack was encouraged to place responsible AI at the core of its business and formulate its own ethical framework to embed in its company culture.
Supported Machine Intelligence Garage, AudioStack raises £1 million in seed funding
Through its early-stage investment series and support, the Machine Intelligence Garage helps to ensure AI startups are investment ready.
In February 2021, AudioStack announced it had raised £1 million in Seed funding from Australian media company Unith (previously Crowd Media). Timo acknowledges that the support from Digital Catapult helped the team to push forward and raise this investment.
“The investment support was superb. For us, it came just at the right time, because we started fundraising last October, which was halfway through the programme,” said Timo Kunz.
Unith took the entire seed round, seeking to integrate AudioStack’s technology with its social commerce application to provide voices for its AI-driven chatbots, with uses that span marketing, education, medical and retail – where influencer avatars can communicate with customers to boost sales.
AudioStack launch AudioStack.ai– a fully automated audio creation environment
In May 2021, AudioStack officially launched AudioStack.ai, an API-based audio production tool for developers. With just a few lines of code, a developer can add text, select a speaker and background track from the library, set the audio speed, then send the final result into production.
To produce work at scale, AudioStack’s powerful rendering engine can generate thousands of different audio versions in seconds from one source file, streamlining the production of versioned audio, such as personalised Spotify or Pandora ads.
In March 2023, AudioStack launched the first ever radio advertising campaign using synthetic media on iHeart radio in the US. The campaign, created with AudioStack.ai ran for a total of 3 weeks in the state of California and reached over 1M listeners.
AudioStack also offers a fully self-serve solution for a generative AI builder. It allows users to create fully-produced audio ads in under 30 seconds by just typing a few words about the product and company they would like to advertise. At the push of a button AudioStack.ai creates inspiring ad copy, chooses the right speaker out of a library of 600 synthetic voices, adds a matching music track and mixes all of these elements into a broadcast-ready audio ad.
“We want to invite people to build innovative audio applications that they couldn’t do before, because production costs were prohibitive or time wouldn’t enable it,” explained Timo. “Being an audio nerd myself, I know that creativity can take many forms and divide people, so I’m very excited to see what people will build when we give them our technology.”
Raised in seed funding from Australian media company Unith
1st version of API audio production engine released in 2021
2nd version with enhanced capabilities released in 2023
Join delivery partners Digital Catapult, Energy Systems Catapult and The Alan Turing Institute for the launch of the Artificial Intelligence…
Helping the innovation community turn theory into practice
Using, extending and creating platforms to accelerate machine learning efforts and generate growth