KRISP audio boot stated on Wednesday that a new feature of AI is starting to change a user’s pronunciation during calls. The company initially circulates support for the change of Indian English intentions to us English.
The start states that the pronunciation process maintains the speaker’s voice and only converts the phonemes that match the American touches. The feature has obviously been tested in business environments and a Beta version is now coming to the Krisp Desktop app. Users can activate the operation at any time during or before calls.
Arto Minasyan, co -founder of the company, said the idea of the characteristic came from a problem he faced in his conversations. “Many people do not understand my pronunciation, even though I speak English well.
However, when this journalist examined the feature, the processed voice does not sound natural and even lost a few words at times. The company attributed these defects to it is a beta release, saying that the model will improve over time.
Krisp claims that when considering business operations, sales conversion rates rose by 26.1% and revenue per book increased by 14.8%.
The company said it has decided to work first in Indian intentions, as people from the country represent a large part of the global workforce in STEM fields. There are plans to add support for more tones, including the Philippines.
Other newly established companies, such as Sanas supported by GV, have developed similar technology in call centers on a scale.
The company says it is training the model in thousands of speech samples that have covered different intentions and dialects and used data from the meeting assistant following the user’s consent.
Minasyan said another advantage of the feature is that no pre-training in a user’s voice as it creates a real-time speaker profile.
KRISP, which last raised in 2021, plans to release iOS and Android applications this year to support personal meetings. There is also a new Chrome extension for better completion with Google Meet in the projects.
