Speechmatics, the Cambridge-based speech technology company, launches its real-time, fully embeddable, continuous speech recognition system in many languages – providing levels of accuracy and speed usually only found on expensive cloud-based services.
Speech recognition technology has been limited to post-processing after an interaction or to the use of short phrases – it was near impossible for large, continuous systems to process fast, accurate transcriptions at scale and over long time periods. Through advances in deep neural networks used in acoustic and language models – Speechmatics have achieved this breakthrough by developing a real-time, speaker independent speech recognition technology, available in many languages. The ground-breaking technology provides device manufacturers and enterprises with the highest level of accuracy and speed for cloud-based or fully embedded services. Speechmatics continuous real-time system has a large 250,000-word vocabulary for each language, optimised for speed and accuracy when deployed on any device. This novel application of RNN’s (a form of deep neural network) in the system allows for it to scale to work for any application, for any use case, in any language – from running on a mobile phone to large server farms. Users are becoming increasingly anxious about data security – Speechmatics’ fully embeddable, real-time system enables the data to be held and processed by the user, running natively on a device, rather than by the cloud. The offline capability takes us one step closer to using speech recognition technology anywhere, at any time, by anyone.
The technology builds on Dr Tony Robinson’s (Founder and chief technology officer of Speechmatics) 30 years of experience in developing speech technologies and writing multiple decoders having already successfully created several speech-tech companies.
In 2016, the Speechmatics R&D team took a step back from the initial focus on developing a novel way to scale RNN language models and completely re-designed the Speechmatics system – with the aim to achieve the same levels of accuracy as the existing batch, cloud-based system. Today, the real-time system achieves the same level of accuracy across all languages due to the innovative use of neural networks. The technology is available on-premises, cloud-based or embedded in a device – enabling immediate, actionable intelligence to be gathered efficiently. The continuous real-time speech recognition system, opens a vast number of use-cases in multiple markets – from instant actionable intelligence in call centres or compliance settings to live sub-titling. From offline email dictation on mobile phones or extending home entertainments from voice commands to true voice interaction.
Dr Hermann Hauser, co-founder of Amadeus Capital Partners, investor in Speechmatics adds:
“We are seeing a shift in the tech industry as we move away from touchpad technology towards speech as the main form of communication. This shift is creating a need for businesses to gain immediate, actionable intelligence through highly accurate speech recognition technology, in many languages. There is strong demand in the market for Speechmatics, as it will allow businesses that work on an international scale to not only ensure speech is transcribed correctly, but also to improve everyday user experiences.” “This is the second big breakthrough we have had within the last 9 months. First, we completed the first version of our Auto-Auto system, which we launched last summer. This was a massive step forward for speech science, as for the first time we could build new languages automatically with 1-2 orders less data and in a matter of weeks. For example, we built Japanese in a few weeks even though no one in the team spoke Japanese. We’re delighted to continue making advances in speech recognition technology, building a strong R&D team and making our achievements globally available.”
comments Speechmatics Chief Technology Officer, Dr Tony Robinson. A free trial demo is now available on the Speechmatics website. Speechmatics will also be demoing the embedded real-time system at MWC 2017, February 28th – March 2nd, Hall 7, stand 7C72.