摘要 |
The present invention discloses methods, apparatus and systems for individualizing music, audio and speech adaptively, intelligently and interactively according to a listener's personal hearing ability, unique hearing preference, characteristic feedback, and real-time surrounding environment. |
主权项 |
1. A sound individualizing system, comprising the steps of: (a) adjusting a sound by an automatic fluctuation control unit; (b) multiplying a plurality of weighting factors with a plurality of data samples of said sound through a sample weighting unit and padding a plurality of zeros by a zero padding unit; (c) transforming a first output of said zero padding unit into a plurality of time-frequency bins by a forward transform unit; (d) passing said time-frequency bins through a cepstrum calculation unit to output a cepstrum; (e) processing said cepstrum by at least one cepstrum-domain lifter; (f) conveying a second output of said lifter into an adaptive classification unit; (g) directing a third output of said forward transform unit to a weighted fusion unit, wherein said weighted fusion unit merges said adjacent time-frequency bins according to human auditory scaling; (h) employing a fourth output of said weighted fusion unit by a long-term moment calculation unit, wherein said long-term moment calculation unit computes a plurality of long-term variance, skewness, kurtosis and higher-order moments; (i) conveying said fourth output of said weighted fusion unit to a short-term moment calculation unit, wherein said short-term moment calculation unit computes a plurality of short-term variance, skewness, kurtosis and higher-order moments; (j) directing said long-term and short-term variances, skewnesses, kurtosises and higher-order moments to said adaptive classification unit; (k) passing said fourth output of said weighted fusion unit to a multi-block weighted averaging unit, wherein said multi-block weighted averaging unit suppresses a plurality of undesired components; (l) calculating a fifth output and a sixth output, wherein said fifth output is a long-term mean value and said sixth output is a short-term mean value; (m) sending said long-term and short-term mean values to said adaptive classification unit, wherein said adaptive classification unit utilizes said cepstrum vector, said long-term and short-term mean values, variances, skewnesses, kurtosises and higher-order moments to classify said sound into a beat category and a non-beat category; (n) converting said beat category and said non-beat category to a beat signal; (o) updating said automatic fluctuation control unit, said sample weighting unit, and said weighting coefficients, wherein said updated weighting coefficients control said multi-block weighted averaging unit to compute said long-term and short-term mean values; and (p) employing said beat signal to enhance auditory perception of an individual listener by an individualized auditory enhancer in accordance to a human input unit. |