Speech recognition method for both english and chinese

The speech recognition method for both English and Chinese provides both English word, sentence, name, and Chinese syllable, sentence, name recognition comprising:

• (1). a pre-processor to delete noise and the time interval without speech signal;
• (2). a method to normalize the whole speech waveform of a known English word (Chinese syllable) using E elastic frames without filter and without overlap and to transform the waveform into an equal-sized E×P matrix of linear predict coding cepstra (LPCC) such that the same English words (Chinese syllables) have about the same LPCC at the same time position in their equal-sized E×P matrices of LPCC;
• (3). in all samples of each known English word (Chinese syllable), select its best K samples to compute the feature to represent the known English word (Chinese syllable);
• (4). the best K samples of a known English word (Chinese syllable) are used to compute the means and all samples of the known English word (Chinese syllable) are used to compute the variances of LPCC and the equal-sized E×P matrix of means and variances is called the standard pattern to represent the feature of the known English word (Chinese syllable) and is saved in the English word and Chinese syllable database;
• (5). a method to normalize the whole waveform of an input unknown English word (Chinese syllable) using E elastic frames without filter and without overlap and to transform the waveform into an equal-sized E×P matrix of linear predict coding cepstra (LPCC), which is called the categorizing pattern to represent the unknown English word (Chinese syllable);
• (6). a simplified Bayesian decision classifier is used to match the standard pattern of every known English word (Chinese syllable) with the categorizing pattern of an input unknown English word (Chinese syllable) and to find a known English word (Chinese syllable) with the least Bayesian distance to be the unknown English word (Chinese syllable);
• (7). a method to modify and improve the feature of an English word (Chinese syllable) such that the English word (Chinese syllable) is guaranteed to be recognized correctly, and a method to create the feature (standard pattern) of a new English word and a new Chinese syllable;
• (8). a method to partition an unknown English (Chinese) sentence or name into D unknown English words (D unknown Chinese syllables);
• (9). a simplified Bayesian classifier to find the F most similar known English words or Chinese syllables or both from the English word and Chinese syllable database for each of D unknown English words (Chinese syllables) and an unknown English (Chinese) sentence or name is represented by a D×F matrix of known similar English words or Chinese syllables or both;
• (10). the D×F matrix of similar known English words (Chinese syllables) matches all known English (Chinese) sentences and names in the English and Chinese sentence and name database and find a known English (Chinese) sentence or name with the highest probability to be the unknown sentence or name;
• (11). a method to modify and improve the feature of an unknown English word (Chinese syllable) in the input unknown English (Chinese) sentence or name such that the input unknown sentence or name is guaranteed to be recognized correctly;
• and
• (12). since the features of English words (without samples) is created from the features of Chinese syllables, the present invention can also create the features of other languages from the features of Chinese syllables and hence the present invention can also recognize many other languages, such as German, French, Japanese, Korean, Russian, etc.