Programmer Guide/Command Reference/EVAL/map2map: Difference between revisions
From STX Wiki
< Programmer Guide | Command Reference | EVAL
Jump to navigationJump to search
No edit summary |
No edit summary |
||
| Line 8: | Line 8: | ||
::<code>2 <= ''nc''[i] <= M; with: i=0..N-1</code> | ::<code>2 <= ''nc''[i] <= M; with: i=0..N-1</code> | ||
;Description: Each row i of the input map contains a data vector {''x''[i,0], .., ''x''[''i,''nc''[i]-1]}. A linear interpolation is used to expand all vectors to the length M. | ;Description: Each row i of the input map contains a data vector {''x''[i,0], .., ''x''[''i,''nc''[i]-1]}. A linear interpolation is used to expand all vectors to the length M. | ||
:Example: ''x'' contains a spectrogram. The spectra are computed with constant hopsize, but different transformation length. The length of each spectrum ''x''[*,i] is stored in ''nc''[i]. | :;Example: ''x'' contains a spectrogram. The spectra are computed with constant hopsize, but different transformation length. The length of each spectrum ''x''[*,i] is stored in ''nc''[i]. | ||
;Result 1:A NxM matrix containing the interpolated map ''y''. | ;Result 1:A NxM matrix containing the interpolated map ''y''. | ||
---- | ---- | ||
| Line 17: | Line 17: | ||
;Description: First the vectors ''x''[i,*] are expanded as described above. In the second step interpolated rows are inserted to create a map with equidistant rows. The number of rows L of the interpolated map ''y'' depends on the range of the ''pr'' values and the minimum distance of two neighboring values of ''pr''. | ;Description: First the vectors ''x''[i,*] are expanded as described above. In the second step interpolated rows are inserted to create a map with equidistant rows. The number of rows L of the interpolated map ''y'' depends on the range of the ''pr'' values and the minimum distance of two neighboring values of ''pr''. | ||
::<code>L = (''pr''[N-1] - ''pr''[0]) / min(''pr''[i] - ''pr''[i-1]) | ::<code>L = (''pr''[N-1] - ''pr''[0]) / min(''pr''[i] - ''pr''[i-1]) | ||
:Example: ''x'' contains a spectrogram. The spectra stored in the columns of ''x'' are computed with varying hopsize and transformation length. The length of each spectrum ''x''[*,i] is stored in ''nc''[i] and the value ''pr''[i] is set to the center positions (or starting time) of the i-th analysis window. | :;Example: ''x'' contains a spectrogram. The spectra stored in the columns of ''x'' are computed with varying hopsize and transformation length. The length of each spectrum ''x''[*,i] is stored in ''nc''[i] and the value ''pr''[i] is set to the center positions (or starting time) of the i-th analysis window. | ||
;Result 2:A LxM matrix containing the interpolated map ''y''. | ;Result 2:A LxM matrix containing the interpolated map ''y''. | ||
---- | ---- | ||
Revision as of 09:59, 12 April 2011
Create a x/y-map with fixed grid (dx and dy is constant) from a x/y-map with varying grid. This function was implemented to display spectrograms with varying hopsize dt and frequency resolution df.
- Usage 1
map2map(type, x, nc)- type
- this argument must be set to 0; it will be used in future to select the format of the input data and/or the remapping algorithm
- x
- a NxM matrix containing the input map
- nc
- a vector with length N; the value nc[i] specifies the number of used columns in the row x[i,*] of the input map
2 <= nc[i] <= M; with: i=0..N-1
- Description
- Each row i of the input map contains a data vector {x[i,0], .., x[i,nc[i]-1]}. A linear interpolation is used to expand all vectors to the length M.
- Example
- x contains a spectrogram. The spectra are computed with constant hopsize, but different transformation length. The length of each spectrum x[*,i] is stored in nc[i].
- Result 1
- A NxM matrix containing the interpolated map y.
- Usage 2
map2map(type, x, nc, pr)- type, x, nc
- see Usage 1
- pr
- a vector with length N; the value pr[i] specifies the relative position of the function stored in the row x[i,*] of the input map.
pr[0] < pr[1] < .. < pr[N-1]
- Description
- First the vectors x[i,*] are expanded as described above. In the second step interpolated rows are inserted to create a map with equidistant rows. The number of rows L of the interpolated map y depends on the range of the pr values and the minimum distance of two neighboring values of pr.
L = (pr[N-1] - pr[0]) / min(pr[i] - pr[i-1])- Example
- x contains a spectrogram. The spectra stored in the columns of x are computed with varying hopsize and transformation length. The length of each spectrum x[*,i] is stored in nc[i] and the value pr[i] is set to the center positions (or starting time) of the i-th analysis window.
- Result 2
- A LxM matrix containing the interpolated map y.
//-----
// Map2Map(Type_s=0, X_m, NCol_v {, PRow_v})
//
// Type_s=0 expand all rows to equal length (use linear interpolation)
// if PRow_v is specified: map varying row distances to fixed grid (use linear interpolation)
//
// X_m input map (N rows, M columns) N > 2, M > 2
// NCol_v NCol_v[n] = the number of used columns in row X_m[n,*]; 0 <= n < N, 2 < NCol[n] < M
// PRow_v XRow_v[n] = the relative position of row X_m[n,*]; 0 <= n < N, PRow_[0] < PRow[1] < .. < PRow[N-2] < PRow[NX-1]
//
// result: without PRow_v: Y_m (N rows, M columns)
// with PRow_v: Y_m (NY rows, M columns); NY := int ( (PRow_v[N-1] - PRow_v[0]) / min(PRow_v) ) + 1
//-----
