For the third time in a row, dawin gmbh made a guest appearance at the Mobile Voice Conference (MVC) in San Jose with high-calibre speakers. Along with Sébastien Bratières, dawin Speech evangelist, this year Filip Korc, dawin Software Architect, spoke about the latest events and practical applications of speech-assisted software solutions.
The MVC is organized by the American AVIOS. AVIOS (Applied Voice Input Output Society) is a private non-profit organisation with the goals of informing, educating and providing resources to developers and designers of new and changing speech technologies.
The Mobile Voice Conference programme can be seen here.
To obtain the presentation material (PDF slides), please leave us a message using our contact form, specifying the talk in which you are interested.
A developer’s tour of affordable options for mobile speech recognition
talk given by dawin Speech Evangelist Sébastien Bratières in the session “Building a conversational application” on Monday, April 20th 2015, 11:00am.
In this talk, we will go through free or nearly-free and open-source ASR services for smartphone apps, among which are PocketSphinx, HTS, Julius, Google Voice, NDEV, iSpeech. We will outline the capabilities of each and give keys to understand their features. We will assess them from the perspective of the speech-technology savvy smartphone app developer.
Available Material:
PDF slides
PDF table of the cloud speech engine providers
Please contact us here
Voice for the Industry Worker
talk given by dawin Software Architect Filip Korc in the session “Personal assistants within the enterprise” on Monday April 20th 2015, 3:15pm
The assessment and recording of plant health in field trials is a key activity in the expensive, time-consuming and risky process of developing an agricultural product. Due to scarcity of experts, this activity is also a real bottleneck. We present a voice-only-driven application enabling parallel hands-free and eyes-free recording to the assessment activity and attacking the bottleneck through raised comfort and effectiveness of the few available experts. Our case study illustrates ample demand for voice in industry processes involving mobile blue-collar workers and in industry applications in general.
Available Material:
PDF slides
Please contact us here
Deep learning in speech recognition: key insights
talk given by dawin Speech Evangelist Sébastien Bratières in the session “A deeper look at the technology” on Monday, April 20th 2015, 11:00am.
While the basic of speech recognition models had not changed since their inception in the 1980’s at IBM in Fred Jelinek’s lab, the surge of performance of deep neural networks in the last three years has reshaped the landscape. Using many examples, we will gently explain what deep neural nets are, how speech recognition used to work before (using Gaussian mixture models), and what techniques are in use today. We will also cover more recent developments such as end-to-end deep neural nets speech recognition which do away with the hidden Markov model, the workhorse of traditional speech recognition, altogether.
Available Material:
PDF slides
Please contact us here