Publisher's Synopsis
The purpose of spectral analysis is to find out how acoustic energy is distributed across frequency. Typical uses in phonetics are discovering the spectral properties of the vowels and consonants of a language, comparing the productions of different speakers, or finding characteristics that point forward to speech perception or back to articulation. The accurate determination of the speech spectrum, particularly for short frames, is commonly pursued in diverse areas including speech processing, recognition, and acoustic phonetics. This book, Speech Spectrum Analysis, makes the subject of spectrum analysis understandable to a wide audience, including those with a solid background in general signal processing and those without such background. The technique of designing musical instruments has not changed much in the last several thousand years. A maker builds an instrument, listens to the tone, then repeats the entire process with a slight change in construction. This is a tedious process and one often thinks that it could be easier if there was a way to "see" the sound. Spectrum analysis is a tool that gives us the ability to see the timbre. To derive spectra from complex sounds we are forced to perform what is called a Fourier transform. The Fourier transform may be visualized as a magic "Black Box" which is able to convert time domain to frequency domain. There are numerous algorithms to accomplish, however the most common is an algorithm known as the "Fast Fourier Transform". The FFT is the most commonly used algorithm for small computer systems. Unfortunately, real world sounds tend to show an absence of such simple repeating patterns. This absence is usually derived from several mechanisms. The first is a random component in the sound. The another is the effect of the envelope (i.e., the attack and decay of the sound). And another deals with different envelopes for each component frequency. Although such fundamental inconsistencies exist between the presumptions of the Fourier transform and the real world, this does not weaken the value of the process. Usually these artifacts are of such a low amplitude that we do not need to worry about them. However, if one suspects that an area of interest may be an artifact, the easiest thing to do is to resample with a different sample size. If the particular component shows wide variation, it is probably an artifact. If it shows a certain consistency then it is probably a legitimate component. Although the Fourier transform may be visualized as "black box" there are still some considerations which should be observed. However, the complexity of the subject still means that there has to be a certain attention to detail. If the nature of sampling and the quirks of the Fourier transform are known, it may be a useful tool for virtually any serious instrument builder, especially with an appropriate graphic output. The purpose of spectral analysis is to find out how acoustic energy is distributed across frequency. Typical uses in phonetics are discovering the spectral properties of the vowels and consonants of a language, comparing the productions of different speakers, or finding characteristics that point forward to speech perception or back to articulation. The accurate determination of the speech spectrum, particularly for short frames, is commonly pursued in diverse areas including speech processing, recognition, and acoustic phonetics. This book, Speech Spectrum Analysis, makes the subject of spectrum analysis understandable to a wide audience, including those with a solid background in general signal processing and those without such background. The technique of designing musical instruments has not changed much in the last several thousand years. A maker builds an instrument, listens to the tone, then repeats the entire process with a slight change in construction. This is a tedious process and one often thinks that it could be easier if there was a way to "see" the sound. Spectrum analysis is a tool that gives us the ability to see the timbre. To derive spectra from complex sounds we are forced to perform what is called a Fourier transform. The Fourier transform may be visualized as a magic "Black Box" which is able to convert time domain to frequency domain. There are numerous algorithms to accomplish, however the most common is an algorithm known as the "Fast Fourier Transform". The FFT is the most commonly used algorithm for small computer systems. Unfortunately, real world sounds tend to show an absence of such simple repeating patterns. This absence is usually derived from several mechanisms. The first is a random component in the sound. The another is the effect of the envelope (i.e., the attack and decay of the sound). And another deals with different envelopes for each component frequency. Although such fundamental inconsistencies exist between the presumptions of the Fourier transform and the real world, this does not weaken the value of the process. Usually these artifacts are of such a low amplitude that we do not need to worry about them. However, if one suspects that an area of interest may be an artifact, the easiest thing to do is to resample with a different sample size. If the particular component shows wide variation, it is probably an artifact. If it shows a certain consistency then it is probably a legitimate component. Although the Fourier transform may be visualized as "black box" there are still some considerations which should be observed. However, the complexity of the subject still means that there has to be a certain attention to detail. If the nature of sampling and the quirks of the Fourier transform are known, it may be a useful tool for virtually any serious instrument builder, especially with an appropriate graphic output. The purpose of spectral analysis is to find out how acoustic energy is distributed across frequency. Typical uses in phonetics are discovering the spectral properties of the vowels and consonants of a language, comparing the productions of different speakers, or finding characteristics that point forward to speech perception or back to articulation. The accurate determination of the speech spectrum, particularly for short frames, is commonly pursued in diverse areas including speech processing, recognition, and acoustic phonetics. This book, Speech Spectrum Analysis, makes the subject of spectrum analysis understandable to a wide audience, including those with a solid background in general signal processing and those without such background. The technique of designing musical instruments has not changed much in the last several thousand years. A maker builds an instrument, listens to the tone, then repeats the entire process with a slight change in construction. This is a tedious process and one often thinks that it could be easier if there was a way to "see" the sound. Spectrum analysis is a tool that gives us the ability to see the timbre. To derive spectra from complex sounds we are forced to perform what is called a Fourier transform. The Fourier transform may be visualized as a magic "Black Box" which is able to convert time domain to frequency domain. There are numerous algorithms to accomplish, however the most common is an algorithm known as the "Fast Fourier Transform". The FFT is the most commonly used algorithm for small computer systems. Unfortunately, real world sounds tend to show an absence of such simple repeating patterns. This absence is usually derived from several mechanisms. The first is a random component in the sound. The another is the effect of the envelope (i.e., the attack and decay of the sound). And another deals with different envelopes for each component frequency. Although such fundamental inconsistencies exist between the presumptions of the Fourier transform and the real world, this does not weaken the value of the process. Usually these artifacts are of such a low amplitude that we do not need to worry about them. However, if one suspects that an area of interest may be an artifact, the easiest thing to do is to resample with a different sample size. If the particular component shows wide variation, it is probably an artifact. If it shows a certain consistency then it is probably a legitimate component. Although the Fourier transform may be visualized as "black box" there are still some considerations which should be observed. However, the complexity of the subject still means that there has to be a certain attention to detail. If the nature of sampling and the quirks of the Fourier transform are known, it may be a useful tool for virtually any serious instrument builder, especially with an appropriate graphic output. The purpose of spectral analysis is to find out how acoustic energy is distributed across frequency. Typical uses in phonetics are discovering the spectral properties of the vowels and consonants of a language, comparing the productions of different speakers, or finding characteristics that point forward to speech perception or back to articulation. The accurate determination of the speech spectrum, particularly for short frames, is commonly pursued in diverse areas including speech processing, recognition, and acoustic phonetics. This book, Speech Spectrum Analysis, makes the subject of spectrum analysis understandable to a wide audience, including those with a solid background in general signal processing and those without such background. The technique of designing musical instruments has not changed much in the last several thousand years. A maker builds an instrument, listens to the tone, then repeats the entire process with a slight change in construction. This is a tedious process and one often thinks that it could be easier if there was a way to "see" the sound. Spectrum analysis is a tool that gives us the ability to see the timbre. To derive spectra from complex sounds we are forced to perform what is called a Fourier transform. The Fourie
