aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorMatthias P. Braendli <matthias.braendli@mpb.li>2016-04-01 10:15:29 +0200
committerMatthias P. Braendli <matthias.braendli@mpb.li>2016-04-01 10:15:29 +0200
commitc5cbfc7e8feb92a65a825698f6c9ca7f6df652db (patch)
treeb602c4897ef86dce2d0d4de633aea7ac2d5739da
parentb04e8d19d684479b4a5c096789ead96f4b1a60ee (diff)
downloadfdk-aac-dabplus-c5cbfc7e8feb92a65a825698f6c9ca7f6df652db.tar.gz
fdk-aac-dabplus-c5cbfc7e8feb92a65a825698f6c9ca7f6df652db.tar.bz2
fdk-aac-dabplus-c5cbfc7e8feb92a65a825698f6c9ca7f6df652db.zip
Fix libtoolame drift compensation
-rw-r--r--src/dabplus-enc.cpp42
1 files changed, 36 insertions, 6 deletions
diff --git a/src/dabplus-enc.cpp b/src/dabplus-enc.cpp
index 5b2df47..34add1c 100644
--- a/src/dabplus-enc.cpp
+++ b/src/dabplus-enc.cpp
@@ -688,12 +688,11 @@ int main(int argc, char *argv[])
/* We assume that we need to call the encoder
* enc_calls_per_output before it gives us one encoded audio
* frame. This information is used when the alsa drift compensation
- * is active
+ * is active. This is only valid for FDK-AAC.
*/
- const int enc_calls_per_output =
- (selected_encoder == encoder_selection_t::fdk_dabplus) ?
- ((aot == AOT_DABPLUS_AAC_LC) ? sample_rate / 8000 : sample_rate / 16000) :
- 1;
+ const int enc_calls_per_output = (aot == AOT_DABPLUS_AAC_LC) ?
+ sample_rate / 8000 :
+ sample_rate / 16000;
int max_size = 8*input_buf.size() + NUM_SAMPLES_PER_CALL;
@@ -819,7 +818,8 @@ int main(int argc, char *argv[])
AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 };
// -------------- wait the right amount of time
- if (drift_compensation || jack_name) {
+ if ( (drift_compensation or jack_name) and
+ selected_encoder == encoder_selection_t::fdk_dabplus) {
struct timespec tp_now;
clock_gettime(CLOCK_MONOTONIC, &tp_now);
@@ -1088,6 +1088,36 @@ int main(int argc, char *argv[])
else {
numOutBytes = toolame_finish(&outbuf[0], outbuf.size());
}
+
+ // We throttle Toolame by measuring the incoming audio size, because we
+ // know it always eats 1152 samples, regardless of configuration.
+ if (read_bytes and (drift_compensation or jack_name)) {
+ struct timespec tp_now;
+ clock_gettime(CLOCK_MONOTONIC, &tp_now);
+
+ unsigned long time_now = (1000000000ul * tp_now.tv_sec) +
+ tp_now.tv_nsec;
+ unsigned long time_next = (1000000000ul * tp_next.tv_sec) +
+ tp_next.tv_nsec;
+
+ // wait_time is in nanoseconds
+ const unsigned long wait_time = 1000000000ul *
+ 1152ul / (unsigned long)sample_rate;
+
+ unsigned long waiting = wait_time - (time_now - time_next);
+ if ((time_now - time_next) < wait_time) {
+ //printf("Sleep %zuus\n", waiting / 1000);
+ usleep(waiting / 1000);
+ }
+
+ // Move our time_counter into the future, for
+ // the next frame.
+ tp_next.tv_nsec += wait_time;
+ if (tp_next.tv_nsec > 1000000000L) {
+ tp_next.tv_nsec -= 1000000000L;
+ tp_next.tv_sec += 1;
+ }
+ }
}
/* Check if the encoder has generated output data */