LiftLabs is launched his own talk model

ElevenA AGAI is just got up a $ 180 million to Meget fundingwas primarily known for their prowness of their generation audio. The company took a step in another technological direction launched his first model of speech in Closdone called scribe.

The startup, rated to $ 3.3 billionhelped many other companies in providing speech services through its victim library. However, the company is now looking to put in detection of speech and compete with the likes of Sword, DeluchMatic, Assembly, On the bottomThe patterns of Openai.

The elevabas’s scribe model will support more than 99 languages ​​to launch. The company hunts more on 25 languages ​​in excellent precision category for the model where the error fare, is less than 5%. This list including English (Boys reporting, French, German, Hindi, Hindi, Japanese Kannadaa, Malailaam, Portuguese, Spanish, Portuguese, Spanish. Other languages ​​are classified categories (5-10% of word words), 10% of the word error of the word), and moderate (25%)

The company said the model used Google Gemini 2.0 flash and mussurraved V3 on multiple languages ​​& you voiced True TrueStump Tests.

The braces have developed the component component in text for its conversational agent platform, which has been released last year. However this is the first time The company is releasing a standalone speech model. I am In a conversation with Techcrough last month, CEE MATI Staniszekski talked about the detection of detection of speech

“We want to understand what it is told by you in a better conversation on the ways to move out of the content only and understanding speech”, staniszekski said at that time. “Many people say the speech-to-text is a problem solved. It is quite that we can build the best of detection of the best speeches because we have a team in-house to ankle data and They give us a quick response. “

The pattern also intelligently to tell you that talk, timestamp to the word level for the sounds sounds sounds and self-tagging as auditory. The Startup provides a way to transcribe clients directly the video content to add subtitles or captions in their study.

The scribe is currently working with pre-recorded audio formats. The company said releasing a low-date version of the model soon. That means that it is not even effective for meetings transcribed or eager-taken voice.

ElevenLabs is to precree scribe to $ 0.40 for an hour of audio transcript. While the fare is competitive, some of their rivals offer a lower price for audio transcriptions at the moment with a characteristic difference.