Senior Research Scientist
Company: Nuance Communications, Inc.
Location: Burlington, MA
Posted on: June 20, 2017
Job Description:
Duties: Develop speech recognition products and solutions,
including Voicemail-to-Text (VM2T) and Voice-to-Text (V2T) solutions and
core speech signaling technologies, such as front-end signal processing
algorithms, Speech Quality Assessment, and Voice Quality Assurance
technologies. Perform V2T research and development to create, deliver, and
improve the core voice-to-text technology components that comprise speech
recognition products. Leverage applied research initiatives of the V2T
group, including the application of VQA technology to the V2T cloud
service, and expansion of Speech Quality Assessment (SQA) technology for use
in VQA. Utilize working knowledge of the VM2T signal processing pipeline
and experience with VQA technologies to expand the SQA technology
for enhanced, future utility outside of VM2T. Develop V2T signal processing
pipeline, SQA, and VQA technology. Research and develop front-end speech
signal processing algorithms for robust automatic speech recognition
and low complexity (compressed domain) speech enhancement, with a
particular focus on VM2T development efforts and compressed domain speech
enhancement. Develop novel speech quality and message classification
algorithms, novel compressed domain speech enhancement algorithms, and robust
speaker segmentation and diarization technology. Build and update models
for QOS service, as well as document current state-of-the-art in speech
quality and related areas of work. Requirements: PhD (or foreign equivalent) in Electrical
Engineering, Signal Processing, or a related field and two (2) years of
experience building and testing signal processing algorithms for speech
recognition products; or in the alternative, a Master’s degree (or foreign
equivalent) in Electrical Engineering, Signal Processing, or a related field
and five (5) years of experience as indicated above. Also requires demonstrated expertise (DE): utilizing Matlab,
Octave and C to simulate degradations in speech processing pipeline and
develop non- intrusive algorithms for CODEC identification, speech quality
and room acoustic parameter estimation (T60, C50, DRR) to optimize and
improve speech processing systems (including speech recognition); DE
performing multi-channel audio signal processing including beamforming and
source tracking for speaker diarization and speech recognition; DE
utilizing speech quality enhancement and noise reduction tools to measure
and improve automatic speech recognition and signal technologies; and DE
leveraging machine learning techniques including speech feature extraction,
feature dimensionality reduction and analysis; and Classification and
Regression modeling with tree based methods (CART), Probabilistic methods
(GMMs) and deep neural networks (DNNs) for speech signal analytics,
speaker diarization and speech recognition. To apply, submit resume referencing Job Code # 31268 to Jason
Hewitt, Nuance Communications, 1 Wayside Road, Burlington, Massachusetts
01803.
Keywords: Nuance Communications, Inc., Haverhill , Senior Research Scientist, Other , Burlington, MA, Massachusetts
Click
here to apply!
|