Vocalis

An AI trained to detect Covid-19 from voice samples

Little by little other new tools based on Artificial Intelligence algorithms are arriving, which try to help in the detection of the presence of Covid-19 or any other potential disease in people, through the analysis of their voice.

In this sense, Vocalis is an emerging company based in Israel, founded in 2019, which is currently working on improving its Artificial Intelligence algorithm for the detection of any potential disease in people, including Covid-19.

Helping to detect Covid-19 where other tools cannot

Vocalis wants to serve where other solutions are ineffective in detecting Covid-19 in people, such as the use of guns to measure their body temperature.

According to TheNextWeb, the algorithm of this company is developed to obtain a total of 512 characteristics of a voice sample, creating a spectrogram about it after which it can detect possible diseases.

The most interesting thing is that the algorithm has been trained to work independently of the language people speak, seeking universal accessibility, for which it has been trained with the voices of a total of 275,000 speakers of different languages, and even has a focus on in the preservation of the identity and privacy of people.

To do this, patients are urged to count only from 50 to 70 so that the voice sample can be drawn for further analysis.

Regarding its effectiveness in the results, during the month of February a research study was carried out in Mumbai, India, with 2,000 people, who were urged to install their VocalisCheck tool on their respective mobiles, concluding that their algorithm had a 81.2% accuracy in the detection of Covid-19, where many of the patients were asymptomatic, and therefore undetectable by other methods.

The company does not want to stop here and try to continue collecting more data that will allow it to continue training its Artificial Intelligence algorithm to be more precise and accurate in its diagnoses.

For now, Vocalis tools are found in the United States, Chile, Indonesia, South Africa, Romania and Luxembourg, both in the public and private sectors, with the idea of ​​being able to commercialize it in more regions.

It will be a matter of time that the algorithm can reach more world regions through the tools of this company, obtaining more precision power over time.