Innovation award winning software turns text into finely-tuned human sounding voices.
Hamilton-based software specialist Company-X has enjoyed international success with text to speech software Voxcoda before launching the Software as a Service (SaaS) solution to the wider market after seeing the potential for future opportunities.
“Voxcoda is a significant opportunity for Company-X and is one of many innovations contributing to our growth in overseas markets,” said Company-X co-founder and director David Hallett.
Company-X built Voxcoda to enable DeLaval in Stockholm, Sweden to create artificial intelligence (AI) generated voices for training video voiceovers, ensuring that the voices sound as human as possible with intricate control over emphasis, pitch, speed and tone.
DeLaval uses e-learning courses to ensure its staff, dealers and millions of farmers in 100 international markets are well-educated in the principles of sustainable food production. Booking voice artists, recording studios and sound engineers for multiple markets was prohibitively expensive.
Voxcoda was built to transform the manual process of creating multiple voice files in the different languages required for DeLaval’s e-learning courses.
The project, initiated because existing solutions did not have the capability DeLaval required, has already won Company-X the Homegrown Innovators Independent Software Vendor category in IDG’s Reseller News Innovation Awards in 2020.
Voxcoda became a pivotal tool for DeLaval Manufacturing milk quality and on-farm service solutions technical development manager Mario Lopez Benavides.
“The flexibility of the Voxcoda tool allows the project team to make sure that voice quality meets the requirements that any user of the final product would expect,” he said.
“Project time is shortened without compromising quality, and that is something we value greatly.”
DeLaval Services farm supplies training and assortment administrator Stefanie Goodhew said: “While working with the recording tool, it is all the more amazing to me how natural the final result sounds and how easily you can change the sound of the words with tiny changes and adjustments.”
Voxcoda uses artificial intelligence (AI) based machine language services with Speech Synthesis Markup Language (SSML) tags to simulate the appropriate accents and languages, with a variety of pitches and tones for each international market, and a combination of male and female voices providing variety.
“The Voxcoda editor has an easy-to-use interface that allows users to easily convert scripts to human-like voice files. The user can enhance the generated voice by adding SSML tags, in the same way that a director might direct an actor,” said Voxcoda product owner Lance Bauerfeind.
“The editor also allows the user to build a pronunciation library of acronyms and domain-specific terminology. The SSML editor delivers text-to-speech services to the DeLaval user at a fraction of the cost of booking a voice artist, recording studio and sound engineer. With this tool, DeLaval has transformed the manual voice translation process internationally into a fully automated digital process.”
Voxcoda is also being used by Stockholm-based multinational CBG to create synthetic audio.
