Phonetic recognition system

Abstract

The invention discloses a speech sound recognition system and includes modulus shift conversion unit, which transfers the analogue signal of the speech sound into the digital signals, feature extraction unit, which carries out the frame treatment to the digital signals and extracts the feature of every frame to get the feature vector sequence, feature codes, which is composed of certain number of codon, quantization enencoding unit, which transfers the feature vector sequence of the input spoken words into the feature codes sequence in according to the feature codes, probability table, which stores probability value of every codon corresponding to the gauss codes in the feature codes and deencoding arithmetic element, which decodes the feature codes sequence to get the recognized results and searches for the gauss codes with the highest matching probability directly in the probability table. The speech sound recognition system can increases the recognition speed of the system on the basis of guaranteeing the performance of the speech sound recognition system.

Description

technical field [0001] The invention relates to a speech recognition system. Background technique [0002] Almost all current speech recognition systems use methods based on statistical pattern recognition. In all speech recognition systems, it is necessary to convert the time-domain sound waves of speech input into a digital vector feature to describe and distinguish different pronunciations, which we call Speech features, based on which a sound model is established for all pronunciations, which is usually called an acoustic model in the field of speech recognition. All speech recognition systems must have an acoustic model; at the same time, for large vocabulary continuous speech recognition systems, a language model is also required. The purpose of speech recognition is to give a string of sound feature sequences as input conditions, use acoustic models and language models, and use search algorithms to output recognition results, such as words, words or sentences. In th...

AI Technical Summary

Problems solved by technology

However, if CDHMM is used in a speech recognition system with a large vocabulary, the Gaussian probability needs to be calculated multiple times when the decoding operation unit performs decoding. Usually, the amount of calculation required in the decoding process is concentrated on the calculation of the Gaussian probability, which requires a lot of Calculations

When large-vocabulary speech recognition is performed on embedded hardware platforms with limited resources such as mobile phones, the response speed of the speech recognition system will be very slow, which cannot meet the needs of actual use.

Images