shithub: opus

Download patch

ref: 80751bad709ce33f892b2f8b34f1e62d20d73574
parent: 4c6a5e0e60f1449c157632b6cf9fecbe663512c3
author: Jean-Marc Valin <jmvalin@amazon.com>
date: Fri Aug 13 22:06:55 EDT 2021 

frame-wise features


--- a/dnn/dump_data.c
+++ b/dnn/dump_data.c
@@ -75,9 +75,9 @@
 }
 
 
-void write_audio(LPCNetEncState *st, const short *pcm, const int *noise, FILE *file) {
+void write_audio(LPCNetEncState *st, const short *pcm, const int *noise, FILE *file, int nframes) {
   int i, k;
-  for (k=0;k<4;k++) {
+  for (k=0;k<nframes;k++) {
   unsigned char data[4*FRAME_SIZE];
   for (i=0;i<FRAME_SIZE;i++) {
     float p=0;
@@ -270,12 +270,15 @@
     if (fpcm) {
         compute_noise(&noisebuf[st->pcount*FRAME_SIZE], noise_std);
     }
+    process_single_frame(st, ffeat);
+    if (fpcm) write_audio(st, pcm, &noisebuf[st->pcount*FRAME_SIZE], fpcm, 1);
+
     st->pcount++;
     /* Running on groups of 4 frames. */
     if (st->pcount == 4) {
-      unsigned char buf[8];
-      process_superframe(st, buf, ffeat, encode, quantize);
-      if (fpcm) write_audio(st, pcmbuf, noisebuf, fpcm);
+      //unsigned char buf[8];
+      //process_superframe(st, buf, ffeat, encode, quantize);
+      //if (fpcm) write_audio(st, pcmbuf, noisebuf, fpcm, 4);
       st->pcount = 0;
     }
     //if (fpcm) fwrite(pcm, sizeof(short), FRAME_SIZE, fpcm);
--- a/dnn/lpcnet_enc.c
+++ b/dnn/lpcnet_enc.c
@@ -860,14 +860,6 @@
   return 0;
 }
 
-void print_vec(float *x, int len) {
-    int i;
-    for (i=0;i<len;i++) {
-        printf("%f ", x[i]);
-    }
-    printf("\n");
-}
-
 LPCNET_EXPORT int lpcnet_compute_features(LPCNetEncState *st, const short *pcm, float features[4][NB_TOTAL_FEATURES]) {
   int i, k;
   for (k=0;k<4;k++) {
@@ -876,13 +868,21 @@
     preemphasis(x, &st->mem_preemph, x, PREEMPHASIS, FRAME_SIZE);
     st->pcount = k;
     compute_frame_features(st, x);
-    process_single_frame(st, NULL);
   }
-  //process_superframe(st, NULL, NULL, 0, 0);
-  //process_multi_frame(st, NULL);
+  process_superframe(st, NULL, NULL, 0, 0);
   for (k=0;k<4;k++) {
     RNN_COPY(&features[k][0], &st->features[k][0], NB_TOTAL_FEATURES);
-    //print_vec(&features[k][0], 20);
   }
+  return 0;
+}
+
+LPCNET_EXPORT int lpcnet_compute_single_frame_features(LPCNetEncState *st, const short *pcm, float features[NB_TOTAL_FEATURES]) {
+  int i;
+  float x[FRAME_SIZE];
+  for (i=0;i<FRAME_SIZE;i++) x[i] = pcm[i];
+  preemphasis(x, &st->mem_preemph, x, PREEMPHASIS, FRAME_SIZE);
+  compute_frame_features(st, x);
+  process_single_frame(st, NULL);
+  RNN_COPY(features, &st->features[0][0], NB_TOTAL_FEATURES);
   return 0;
 }
-- 
⑨