Speaker 1: Can you imagine what we can do today with a speech processing system that is completely automatic, that identifies the spoken language, that recognizes who is speaking, and that writes everything that is said? We, yes. Until now, the performance of speech processing systems did not allow to go beyond the search for keywords. Recent progress made in recent years, especially within the framework of Quero, now allows to consider more and more ambitious applications such as automatic teleconferencing transcription. Let's take an example. During the presentation, all participants will be in a listening mode. Assuming these recent corn prices hold, the 170 recent negative impact in 2013 will automatically be converted into a raw text, which is then automatically enriched by the addition of punctuation, breakage, and writing of numbers and dates. Our partner then performs the only non-automatic processing of the chain. The rereading and correction by a human to publish a verified transcription without error. Finally, these transcriptions are published on the client's website, 15 minutes after the end of each teleconference. At the end of each campaign, the manually corrected data is returned to Vocapia. We use them to adapt our models and thus reduce the number of transcription errors for the next campaign, thereby reducing manual correction costs and therefore production costs. The number of teleconferences processed tripled in a year. The commercialization of our speech transcription software via a web service is a great success. The VoxSigma software suite is available for 17 languages, soon to be 25. The automatic transcription of speech is used to reduce both the time and production costs of transcriptions. It also allows to enhance audio data by allowing them to benefit from tools currently available for text. Our software is at the forefront of technology. In addition, we offer solutions tailored to the specific needs of our clients to maximize their return on investment.