International Business Machines Corporation

Lsi Logic Corporation

Lsi Corporation

Other

Nagendra Kumar Goel

Ramesh Ambat Gopinath

Sabine V Deligne

Ran Katzur

Mousmita Sarma

A real-time dialogue system that provides real-time transcription of the spoken text, with a sub-second delay by keeping track of word timings and word accuracy is provided. The system uses a grammar or a list of keywords to produce the transcripts by using a statistical language model. In addition, the system uses a deep neural network based I-vector system to constantly analyze the audio quality to assess and to identify additional metadata such as gender, language, accent, age, emotion and identity of an end user to enhance the response. The present invention provides a conversational dialogue system, to robustly identify certain specific user commands or intents, while otherwise allowing for a natural conversation, without switching between grammar based and natural language modes.

Dialogue system incorporating unique speech to text conversion method for meaningful dialogue response

A method and apparatus which provides that, in a voice over network, incoming packets are analyzed and the appropriate moment to increase or decrease the nominal delay associated with a jitter buffer is determined. Hence, the nominal delay is adjusted at an appropriate moment based on network jitter characteristics. Preferably, the nominal delay is adjusted when voice activity is absent. The method and apparatus provide for improved play out of the jitter buffer, and provide improved performance.

Method and apparatus for improved play-out packet control algorithm

In contrast to prior are solutions that conduct an averaging operation to estimate metrics related to a communications channel such as a signal-to-noise ratio, the present invention selects a particular value such as a minimum, maximum, or median value from a distribution of values collected over a selected interval. Selecting a particular value from the distribution of values facilitates a more accurate characterization process and increased data throughput. To reduce the processing burden associated with selecting a particular value, the present invention provides a set of cascaded value selection queues that each selects a particular value from the queued values such as a minimum value. The cascaded queues are also successively sub-sampled to reduce the computing resources required to characterize the communications channel. The estimated metrics resulting from the above-described process may be used to adjust the data encoding process and increase data throughput on the communications channel.

Apparatus method and system for characterizing a communication channel and increasing data throughput

A method and apparatus for using and combining sub-frame processing and adaptive jitter-buffers for improved voice quality in voice-over-packet networks. Data is placed in a jitter buffer, where the data has a frame-length consisting of a plurality of samples. Some of the samples are placed in the DMA buffer, and some of the samples are placed in the back-up buffer. Samples are read out of the DMA buffer, and samples are moved from the back-up buffer to the DMA buffer.

Method and apparatus for using and combining sub-frame processing and adaptive jitter-buffers for improved voice quality in voice-over-packet networks

Noise and signal-to-noise ratio (SNR) estimation are relatively straightforward tasks. However, when SNR is small, systematic errors in measurement may result in over-estimation of SNR, which also occurs during runtime monitoring of SNR. Here, sufficient numbers of bits have been preassigned to each channel using QAM modulation scheme. Therefore, SNR relative to QAM lattice size depends on the noise margin and the desired (bit error rate) BER. If a relatively small margin is desired, similar measurement errors may result in over-estimation of SNR. Another problem that arises is that the variance of the noise estimator is relatively high. Therefore, SNR estimates may vary by several dB, and there is only 50% confidence in the usual estimators that the actual SNR value will not be worse than that estimated. Thus, a computationally efficient method for SNR estimation that also allows for specification of a confidence level in the estimates is provided.

Method for accurate estimation of noise for data modems

A system and method for applying a linear transformation to classify and input event. In one aspect, a method for classification comprises the steps of capturing an input event; extracting an n-dimensional feature vector from the input event; applying a linear transformation to the feature vector to generate a pool of projections; utilizing different subsets from the pool of projections to classify the feature vector; and outputting a class identity of the classified feature vector. In another aspect, the step of utilizing different subsets from the pool of projections to classify the feature vector comprises the steps of, for each predefined class, selecting a subset from the pool of projections associated with the class; computing a score for the class based on the associated subset; and assigning, to the feature vector, the class having the highest computed score.

System and method for enhancing speech and pattern recognition using multiple transforms

Methods and apparatus for enhancing the accuracy and efficacy of speech recognition in an environment, such as the interior of an automobile or home, in which there may be significant ambient noise, such as music from loudspeakers within an automobile or programming from a television within a home.

Growing community of inventors

McLean, VA, United States of America

Nagendra Kumar Goel