1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
|
;; spectral-analysis.lsp -- functions to simplify computing
;; spectrogram data
;;
;; Roger B. Dannenberg and Gus Xia
;; Jan 2013, modified Oct 2017
;; API:
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; set sa-obj = sa-init(resolution: <nil or Hz>,
;; fft-dur: <nil or seconds>,
;; skip-period: <seconds>,
;; window: <window type>,
;; input: <filename or sound>)
;;
;; sa-init() creates a spectral-analysis object that can be used
;; to obtain spectral data from a sound.
;;
;; resolution is the width of each spectral bin in Hz. If nil of
;; not specified, the resolution is computed from fft-dur.
;; The actual resolution will be finer than the specified
;; resolution because fft sizes are rounded to a power of 2.
;; fft-dur is the width of the FFT window in seconds. The actual
;; FFT size will be rounded up to the nearest power of two
;; in samples. If nil, fft-dur will be calculated from
;; resolution. If both fft-size and resolution are nil
;; or not specified, the default value of 1024 samples,
;; corresponding to a duration of 1024 / signal-sample-rate,
;; will be used. If both resolution and fft-dur are
;; specified, the resolution parameter will be ignored.
;; Note that fft-dur and resolution are reciprocals.
;; skip-period specifies the time interval in seconds between
;; successive spectra (FFT windows). Overlapping FFTs are
;; possible. The default value overlaps windows by 50%.
;; Non-overlapped and widely spaced windows that ignore
;; samples by skipping over them entirely are also acceptable.
;; window specifies the type of window. The default is raised
;; cosine (Hann or "Hanning") window. Options include
;; :hann, :hanning, :hamming, :none, nil, where :none and
;; nil mean a rectangular window.
;; input can be a string (which specifies a sound file to read)
;; or a Nyquist SOUND to be analyzed.
;; Return value is an XLISP object that can be called to obtain
;; parameters as well as a sequence of spectral frames.
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; set sa-frame = sa-next(sa-obj)
;;
;; sa-next() fetches the next spectrum from sa-obj.
;;
;; sa-obj is a spectral-analysis object returned by sa-init().
;; Return value is an array of FLONUMS representing the discrete
;; spectrum.
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; exec sa-info(sa-obj)
;;
;; sa-info prints information about the spectral computation.
;;
;; sa-obj is a spectral-analysis object returned by sa-init().
;; Return value is nil, but information is printed.
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; set mag = sa-magnitude(frame)
;;
;; sa-magnitude computes the magnitude (amplitude) spectrum
;; from a frame returned by sa-frame.
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; exec sa-plot(sa-obj, sa-frame)
;;
;; sa-plot plots the amplitude (magnitude) spectrum of sa-frame.
;;
;; sa-obj is used to determine the bin width of data in sa-frame.
;;
;; sa-frame is a spectral frame (array) returned by sa-next()
;;
;; Return value is nil, but a plot is generated and displayed.
;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; set hz = sa-get-bin-width(sa-obj)
;; set n = sa-get-fft-size(sa-obj)
;; set secs = sa-get-fft-dur(sa-obj)
;; set window = sa-get-fft-window(sa-obj)
;; set skip-period = sa-get-skip-period(sa-obj)
;; set m = sa-get-fft-skip-size(sa-obj)
;; set sr = sa-get-sample-rate(sa-obj)
;;
;; These functions retrieve data from the sa-obj created by
;; sa-init. The return values are:
;; hz - the width of a frequency bin (also the separation
;; of bin center frequencies). The center frequency of
;; the i'th bin is i * hz.
;; n - the size of the FFT, an integer, a power of two. The
;; size of a spectral frame (an array returned by sa-next)
;; is (n / 2) + 1.
;; secs - the duration of an FFT window.
;; window - the type of window used (:hann, :hamming, :none)
;; skip-period - the time in seconds of the skip (the time
;; difference between successive frames
;; m - the size of the skip in samples.
;; sr - the sample rate of the sound being analyzed (in Hz, a flonum)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; define the class of spectral analysis objects
(setf sa-class (send class :new '(sound length skip window window-type)))
(send sa-class :answer :next '() '(
(snd-fft sound length skip window)))
(defun sa-raised-cosine (alpha beta)
(sum (const alpha)
(scale beta (lfo 1.0 1.0 *sine-table* 270))))
(defun sa-fft-window (frame-size alpha beta)
(abs-env (control-srate-abs frame-size
(sa-raised-cosine alpha beta))))
(defun hann-window (frame-size) (sa-fft-window frame-size 0.5 0.5))
(defun hamming-window (frame-size) (sa-fft-window frame-size 0.54 0.46))
(defun sa-get-window-type (win-type)
(case win-type
((:hann :hanning) :hann)
((nil :none) :none)
(:hamming :hamming)
(t (print "Warning: invalid window-type parameter: ~A~%" win-type)
(print " Using :HAMMING instead.~%")
:hamming)))
(defun sa-compute-window (len win-type)
(case win-type
(:hann (hann-window len))
(:none nil)
(:hamming (hamming-window len))
(t (print "Warning: invalid window-type parameter: ~A~%" win-type)
(print " Using :HAMMING instead.~%")
(hamming-window len))))
(send sa-class :answer :isnew '(snd len skp win-type) '(
(setf sound snd)
(setf length len)
(setf skip skp)
(setf window-type (sa-get-window-type win-type))
(setf window (sa-compute-window length window-type))))
;; sa-to-mono -- sum up the channels in an array
;;
(defun sa-to-mono (s)
(let ((mono (aref s 0)))
(dotimes (i (1- (length s)))
(setf mono (sum mono (aref s (1+ i)))))
mono))
(defun sa-init (&key resolution fft-dur skip-period window input)
(let (len sr n skip)
(cond ((stringp input)
(setf input (s-read input))))
(cond ((arrayp input)
(format t "Warning: sa-init is converting stereo sound to mono~%")
(setf input (sa-to-mono input)))
((soundp input) ;; so that variables are not "consumed" by snd-fft
(setf input (snd-copy input))))
(cond ((not (soundp input))
(error
(format nil
"Error: sa-init did not get a valid :input parameter~%"))))
(setf sr (snd-srate input))
(setf len 1024)
(cond (fft-dur
(setf len (* fft-dur sr)))
(resolution
(setf len (/ sr resolution))))
;; limit fft size to between 4 and 2^16
(cond ((> len 65536)
(format t "Warning: fft-size reduced from ~A to 65536~%" len)
(setf len 65536))
((< len 4)
(format t "Warning: fft-size increased from ~A to 4~%" len)
(setf len 4)))
;; round up len to a power of two
(setf n 4)
(while (< n len)
(setf n (* n 2)))
(setf length n) ;; len is now an integer power of 2
;(display "sa-init" length)
;; compute skip length - default is len/2
(setf skip (if skip-period (round (* skip-period sr))
(/ length 2)))
(send sa-class :new input length skip window)))
(defun sa-next (sa-obj)
(send sa-obj :next))
(defun sa-info (sa-obj)
(send sa-obj :info))
(send sa-class :answer :info '() '(
(format t "Spectral Analysis object (instance of sa-class):~%")
(format t " resolution (bin width): ~A Hz~%" (/ (snd-srate sound) length))
(format t " fft-dur: ~A s (~A samples)~%" (/ length (snd-srate sound)) length)
(format t " skip-period: ~A s (~A samples)~%" (/ skip (snd-srate sound)) skip)
(format t " window: ~A~%" window-type)
nil))
(defun sa-plot (sa-obj frame)
(send sa-obj :plot frame))
(defun sa-magnitude(frame)
(let* ((flen (length frame))
(n (/ (length frame) 2)) ; size of amplitude spectrum - 1
(as (make-array (1+ n)))) ; amplitude spectrum
;; first compute an amplitude spectrum
(setf (aref as 0) (abs (aref frame 0))) ;; DC
;; half_n is actually length/2 - 1, the number of complex pairs
;; in addition there is the DC and Nyquist terms, which are
;; real and in the first and last slots of frame
(setf half_n (1- n))
(dotimes (i half_n)
(let* ((i2 (+ i i 2)) ; index of the imag part
(i2m1 (1- i2)) ; index of the real part
(amp (sqrt (+ (* (aref frame i2m1) (aref frame i2m1))
(* (aref frame i2) (aref frame i2))))))
(setf (aref as (1+ i)) amp)))
(setf (aref as n) (aref frame (1- flen)))
as)) ;; return the amplitude spectrum
(send sa-class :answer :plot '(frame) '(
(let* ((as (sa-magnitude frame))
(sr (snd-srate sound)))
(s-plot (snd-from-array 0 (/ length sr) as)
sr (length as)))))
(defun sa-get-bin-width (sa-obj)
(send sa-obj :get-bin-width))
(send sa-class :answer :get-bin-width '()
'((/ (snd-srate sound) length)))
(defun sa-get-fft-size (sa-obj)
(send sa-obj :get-fft-size))
(send sa-class :answer :get-fft-size '() '(length))
(defun sa-get-fft-dur (sa-obj)
(send sa-obj :get-fft-dur))
(send sa-class :answer :get-fft-dur '() '(/ length (snd-srate sound)))
(defun sa-get-fft-window (sa-obj)
(send sa-obj :get-fft-window))
(send sa-class :answer :get-fft-window '() '(window-type))
(defun sa-get-fft-skip-period (sa-obj)
(send sa-obj :get-skip-period))
(send sa-class :answer :get-skip-period '() '((/ skip (snd-srate sound))))
(defun sa-get-fft-skip-size (sa-obj)
(send sa-obj :get-skip-size))
(send sa-class :answer :get-fft-skip-size '() '(skip))
(defun sa-get-sample-rate (sa-obj)
(send sa-obj :get-sample-rate))
(send sa-class :answer :get-sample-rate '() '((snd-srate sound)))
;;;;;;; TESTS ;;;;;;;;;;
(defun plot-test ()
(let (frame)
(setf sa (sa-init :input "./rpd-cello.wav"))
(while t
(setf frame (sa-next sa))
(if (null sa) (return nil))
(sa-plot sa frame))))
|