Changeset 7100895

Timestamp:

Jul 29, 2017, 10:11:52 PM (7 years ago)

Author:

Eduard Müller <mueller.eduard@googlemail.com>

Branches:

feature/autosink, feature/cnn, feature/cnn_org, feature/constantq, feature/crepe, feature/crepe_org, feature/pitchshift, feature/pydocstrings, feature/timestretch, fix/ffmpeg5, master

Children:

faeec7c

Parents:

986131d

Message:

Revert ooura reverse fft scaling changes

... and adjusted new Intel IPP fft impls scaling accordingly as well

File:

• src/spectral/fft.c (modified) (1 diff)

src/spectral/fft.c

@@ -450 +450 @@
   ippsFFTInv_CCSToR_64f((const Ipp64f *)s->complexOut, output->data, s->fftSpec, s->memBuffer);
   // apply scaling
-  ippsMulC_64f(output->data, 1.0 / s->winsize, output->data, s->fft_size);
+  ippsMulC_64f(output->data, 2.0 / s->winsize, output->data, s->fft_size);
 #else
   // apply fft
   ippsFFTInv_CCSToR_32f((const Ipp32f *)s->complexOut, output->data, s->fftSpec, s->memBuffer);
   // apply scaling
-  ippsMulC_32f(output->data, 1.0f / s->winsize, output->data, s->fft_size);
+  ippsMulC_32f(output->data, 2.0f / s->winsize, output->data, s->fft_size);
 #endif /* HAVE_AUBIO_DOUBLE */
 
 #else                         // using OOURA
-  smpl_t scale = 1.0 / s->winsize;
+  smpl_t scale = 2.0 / s->winsize;
   s->out[0] = compspec->data[0];
   s->out[1] = compspec->data[s->winsize / 2];