< >
< >
< >
< >
< >
In overcrowded VoiceKI - current-scope.com
< >
< >

In overcrowded VoiceKI


Would you like to insight in your inbox? Register for our weekly newsletters to only receive the company manager of Enterprise AI, data and security managers. Subscribe now


Openai contributes to an increasingly competitive AI language market for companies with its New model, GPT-RealimeThis follows complex instructions and with voices that “sound natural and expressive”.

As a language ski continues to grow and Customers find applications How customer calls or real-time translations, the market for realistically sounding AI voices, which also offer security for corporate quality, heats up. Openaai claims that his new model is a more human voice, but it still has to compete against companies such as Elevenlabs.

The model will be Available on the real time -apiWhat the company has also made in general. Together with the GPT realtime model, Openai also published new voices on the API, which it calls Cedar and Marin, and updated his other voices to work with the latest model.

Openaai said in a live stream that it worked with its customers who built voice applications for the training of GPT-Realime and “carefully geared the model to evals that are based on real scenarios such as customer support and academic tutoring”.


AI scale hits its limits

Power caps, rising token costs and infection delays change the company -ai. Take our exclusive salon to find out how top teams: Top teams are:

  • Transform energy into a strategic advantage
  • Architects efficient inference for real throughput gains
  • Development of the competition -roi with sustainable AI systems

Secure your place to stay in front: https://bit.ly/4mwgngo


https://www.youtube.com/watch?v=nfbbmmjhx0

The company has advertised the ability of the model to create emotional, naturally sounding voices that also match the structure of developers.

Language-to-language models

The model works within a language-to-speech frameworks so that it can understand and react spoken. Language-to-language models are ideal for real-time words in which a person, usually a customer, interacts with an application.

For example, a customer would like to return some products and call a customer service platform. You could speak to a AI language assistant who responds to questions and inquiries as if you are talking to a person.

In a livestream, Openai customer T-mobile Presented a AI speaker who helps people find new phones. Another customer, the real estate search platform ZillowPresented an agent who helps someone narrow a neighborhood to find the perfect place.

Openaai said GPT-Realime was the “most advanced, ready-to-production language model”. Like its other language models, the languages ​​can change the languages ​​in the middle of the legend. However, Openai researchers found that GPT-Realime can follow more complex instructions such as “speaking in a French accent”.

However, the GPT realtime is pending on the competition of other models that many brands already use. Elfflabs released Conversation AI 2.0 in May. Soundhound Partner with fast food franchise company for a Ki-Voice-drive-Thru. Emphatic AI startup Human has Started his EVI 3 modelWhat enables users to generate AI versions of their own voice.

While companies discover various applications for language skis, even more general model providers that offer multimodal LLMs make a case in themselves. mistral published his new one Voxtral modelIt would work well with real -time translation. Google reinforced His audio functions and achieve with a popularity Audio function on NotboKlm This converts research notes into a podcast.

Better instructions follow

Openaai said that GPT realtime is smarter and understands a native audio better, including the ability to catch non-verbal information such as laughter or sigh.

The benchmarking using the Big Bench Audio Eval showed the model, which rated the accuracy of 82.8%, compared to its predecessor, which achieved 65.6%. Openai did not provide any numbers that GPT-Realime tested against models from its competitors.

Openai focused on improving the instruction functions of the model and ensuring that the model would comply with effective instructions. The new model achieves a score of 30.5% on the multichals audio benchmark. The engineers have also called up functions so that GPT realtime can access the right tools.

Real -time -api updates

In order to support and improve the new model how companies integrate real-time AI functions into their applications, Openai has added several new functions to the real-time API.

It can now support MCP and recognize image entries so that it informs users about what it sees in real time. This is a function that Google emphasized during it Project Astra Presentation last year.

The real -time -API can also process the session protocol (SIP). SIP combines apps with telephones such as a public telephone network or a desk telephones, whereby further applications of contact center are opened. Users can also save and reuse input requests on the API.

So far, people have been impressed by the model, although these are still the first tests of a recently published model.

Openaai reduced prices for GPT realtime by 20% to $ 32 per million audio input tokens and $ 64 for audio output tokens.


Leave a Reply

Your email address will not be published. Required fields are marked *

< >