QUICKSPEECH Project

The QuickSpeech project aims to create a new research hub within VoiceInteraction that will aim to research and develop new models, algorithms and/or software architectures to optimize our language models’ processing.

Description:
This is a hub to support the basic processes that operate at the root of the technologies that VoiceInteraction uses in its solutions and, therefore, represents an important strategic bet for it.
Due to this need, VoiceInteraction proposes to design a research center that will aim to investigate and test:
  • New algorithms for parallel data distribution and processing via software;
  • New data distribution and processing algorithms using the parallelization of GPUs (graphic processing units), FPGAs (Field-Programmable Gate Arrays) or ASICs (Application Specific Integrated Circuits);
  • Methods to make the developed models hardware agnostic (ie, integrable with any type of data processing hardware - new GPU technologies, FPGAs, ASICs, new processing hardware, new architectures for parallel data distribution, etc.);
  • Methods to make the developed models agnostic to the base technology (that is, they can be applied to make the processing of Deep Learning algorithms, Neural Networks, Machine Learning, or any other type of algorithms that require a high data processing load more efficient)

Ideally, the actions carried out by this research center culminate in the development of a software middleware that is capable of:
  • Identify available hardware processing resources;
  • Interpret the speech processing algorithms that will process;
  • Understand the processing actions of this algorithm;
  • Distribute processing actions across the various available hardware.

The formation of the proposed hub is a strategic stake by the company to ensure that its speech processing algorithms maintain the effectiveness and speed of execution both in the present and in the future. This is because it is expected that data sources will accommodate an increasing number of data and that processing algorithms and technologies will demand more and more processing resources. It also allows VoiceInteraction to develop new solutions based on more intelligent speech processing algorithms that can be integrated into new areas and new markets.


Grants:
Project name: QuickSpeech

Project code: LISBOA-01-0247-FEDER-042679

Main Objective: Strengthen research, technological development and innovation

Intervention region: Lisbon

Beneficiary entity: VOICEINTERACTION - TECNOLOGIAS DE PROCESSAMENTO DA FALA S.A.

Approval date: 2019-03-28

Start date: 01-02-2019

Completion date: 2021-01-31

Total eligible cost: €263.471,87

Financial support from the European Union | FEDER: €105.388,75


Call us today at +1 646 504 7906 or Email us at info@voiceinteraction.tv

Stay in touch

Request a demo and experience what our Speech Processing platforms can offer you.

Call us today at
+1 646 504 7906
or Email us at info@voiceinteraction.tv

Request a demo and experience what our Speech Processing platforms can offer you.

Stay in touch