Speed and accuracy to make audio talk!
Near real-time transcription (streaming)
Access to full text and metadata for full-text indexing
Detection of key words or phrases
Direct access to desired content with word-level time coding
Daily update of language model and dictionary
An engine boosted by deep learning
Thanks to Deep Neural Networks, MediaSpeech® relies on an extremely fine modeling of the acoustic space, robust to the variability of speakers (flow, accent, …) and sound conditions, and thus offers an even more accurate transcription.
Who is speaking, when, and in which language?
- Time-coded segmentation (speech turns) into speakers, with gender recognition
- Speaker identification via a biometric database
- Language identification and multilingual transcription: French English (US/UK) Arab Spanish Italy Flemish German Russian Chinese (Mandarin)
Available soon : Japanese, Korean and Cantonese Chinese
Ministry of tourism saudi Arabia
MediaSpeech® technology is deployed together with a partner indexing/search technology to enhance the Saudi tourism portal multimedia experience.
Media monitoring leader companies in UK & Spain
MediaSpeech® usage is deeply integrated in the Customer functional workflow for operational excellence on high volume of radio and TV programs.
Online press leader in France
MediaSpeech® technology is integrated into the Media Asset Management system to help journalists exploit video sources.
Turnkey solution, easy to integrate into your information systems to give them a Speech functionality.
High-availability cluster solution for largescale processing of critical infrastructure data.
MediaSpeech® Virtual Machine
Turnkey virtual software solution.
“Software-as-a-service” solution on a pay-as-you-go basis, hosted in Bertin IT’s private cloud.