ardour/libs/ardour/session_ltc.cc

678 lines
22 KiB
C++

/*
* Copyright (C) 2012-2017 Robin Gareus <robin@gareus.org>
* Copyright (C) 2012-2019 Paul Davis <paul@linuxaudiosystems.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include "temporal/time.h"
#include "ardour/audioengine.h"
#include "ardour/audio_port.h"
#include "ardour/debug.h"
#include "ardour/io.h"
#include "ardour/session.h"
#include "ardour/transport_master.h"
#include "ardour/transport_master_manager.h"
#include "pbd/i18n.h"
using namespace std;
using namespace ARDOUR;
using namespace PBD;
using namespace Timecode;
/* really verbose timing debug */
//#define LTC_GEN_FRAMEDBUG
//#define LTC_GEN_TXDBUG
#ifndef MAX
#define MAX(a,b) ( (a) > (b) ? (a) : (b) )
#endif
#ifndef MIN
#define MIN(a,b) ( (a) < (b) ? (a) : (b) )
#endif
/* LTC signal should have a rise time of 25 us +/- 5 us.
* yet with most sound-cards a square-wave of 1-2 sample
* introduces ringing and small oscillations.
* https://en.wikipedia.org/wiki/Gibbs_phenomenon
* A low-pass filter in libltc can reduce this at
* the cost of being slightly out of spec WRT to rise-time.
*
* This filter is adaptive so that fast vari-speed signals
* will not be affected by it.
*/
#define LTC_RISE_TIME(speed) MIN (100, MAX(40, (4000000 / ((speed==0)?1:speed) / engine().sample_rate())))
#define TV_STANDARD(tcf) \
(timecode_to_frames_per_second(tcf)==25.0 ? LTC_TV_625_50 : \
timecode_has_drop_frames(tcf)? LTC_TV_525_60 : LTC_TV_FILM_24)
void
Session::ltc_tx_initialize()
{
assert (!ltc_encoder && !ltc_enc_buf);
ltc_enc_tcformat = config.get_timecode_format();
ltc_tx_parse_offset();
DEBUG_TRACE (DEBUG::TXLTC, string_compose("LTC TX init sr: %1 fps: %2\n", nominal_sample_rate(), timecode_to_frames_per_second(ltc_enc_tcformat)));
ltc_encoder = ltc_encoder_create(nominal_sample_rate(),
timecode_to_frames_per_second(ltc_enc_tcformat),
TV_STANDARD(ltc_enc_tcformat), 0);
ltc_encoder_set_bufsize(ltc_encoder, nominal_sample_rate(), 23.0);
ltc_encoder_set_filter(ltc_encoder, LTC_RISE_TIME(1.0));
/* buffersize for 1 LTC sample: (1 + sample-rate / fps) bytes
* usually returned by ltc_encoder_get_buffersize(encoder)
*
* since the fps can change and A3's min fps: 24000/1001 */
ltc_enc_buf = (ltcsnd_sample_t*) calloc((nominal_sample_rate() / 23), sizeof(ltcsnd_sample_t));
ltc_speed = 0;
ltc_prev_cycle = -1;
ltc_tx_reset();
ltc_tx_resync_latency (true);
Xrun.connect_same_thread (ltc_tx_connections, boost::bind (&Session::ltc_tx_reset, this));
LatencyUpdated.connect_same_thread (ltc_tx_connections, boost::bind (&Session::ltc_tx_resync_latency, this, _1));
restarting = false;
}
void
Session::ltc_tx_cleanup()
{
DEBUG_TRACE (DEBUG::TXLTC, "cleanup\n");
ltc_tx_connections.drop_connections ();
free(ltc_enc_buf);
ltc_enc_buf = NULL;
ltc_encoder_free(ltc_encoder);
ltc_encoder = NULL;
}
void
Session::ltc_tx_resync_latency (bool playback)
{
if (deletion_in_progress() || !playback) {
return;
}
std::shared_ptr<Port> ltcport = ltc_output_port();
if (ltcport) {
ltcport->get_connected_latency_range(ltc_out_latency, true);
DEBUG_TRACE (DEBUG::TXLTC, string_compose ("resync latency: %1\n", ltc_out_latency.max));
}
}
void
Session::ltc_tx_reset()
{
DEBUG_TRACE (DEBUG::TXLTC, "reset\n");
assert (ltc_encoder);
ltc_enc_pos = -9999; // force re-start
ltc_buf_len = 0;
ltc_buf_off = 0;
ltc_enc_byte = 0;
ltc_enc_cnt = 0;
ltc_encoder_reset(ltc_encoder);
}
void
Session::ltc_tx_parse_offset() {
Timecode::Time offset_tc;
Timecode::parse_timecode_format(config.get_timecode_generator_offset(), offset_tc);
offset_tc.rate = timecode_frames_per_second();
offset_tc.drop = timecode_drop_frames();
timecode_to_sample(offset_tc, ltc_timecode_offset, false, false);
ltc_timecode_negative_offset = !offset_tc.negative;
ltc_prev_cycle = -1;
}
void
Session::ltc_tx_recalculate_position()
{
SMPTETimecode enctc;
Timecode::Time a3tc;
ltc_encoder_get_timecode(ltc_encoder, &enctc);
a3tc.hours = enctc.hours;
a3tc.minutes = enctc.mins;
a3tc.seconds = enctc.secs;
a3tc.frames = enctc.frame;
a3tc.rate = timecode_to_frames_per_second(ltc_enc_tcformat);
a3tc.drop = timecode_has_drop_frames(ltc_enc_tcformat);
Timecode::timecode_to_sample (a3tc, ltc_enc_pos, true, false,
(double)sample_rate(),
config.get_subframes_per_frame(),
ltc_timecode_negative_offset, ltc_timecode_offset
);
restarting = false;
}
void
Session::send_ltc_for_cycle (samplepos_t start_sample, samplepos_t end_sample, pframes_t n_samples)
{
assert (n_samples > 0);
pframes_t txf = 0;
std::shared_ptr<Port> ltcport = ltc_output_port();
if (!ltcport) {
assert (deletion_in_progress ());
return;
}
Buffer& buf (ltcport->get_buffer (n_samples));
buf.silence (n_samples);
if (!ltc_encoder || !ltc_enc_buf) {
return;
}
if (!TransportMasterManager::instance().current()) {
return;
}
SyncSource sync_src = TransportMasterManager::instance().current()->type();
if (engine().freewheeling() || !Config->get_send_ltc()
/* TODO
* decide which time-sources we can generated LTC from.
* Internal, JACK or sample-synced slaves should be fine.
* talk to oofus.
*
|| (config.get_external_sync() && sync_src == LTC)
|| (config.get_external_sync() && sync_src == MTC)
*/
||(config.get_external_sync() && sync_src == MIDIClock)
) {
return;
}
Sample* out = dynamic_cast<AudioBuffer*>(&buf)->data ();
/* range from libltc (38..218) || - 128.0 -> (-90..90) */
const float ltcvol = Config->get_ltc_output_volume()/(90.0); // pow(10, db/20.0)/(90.0);
DEBUG_TRACE (DEBUG::TXLTC, string_compose("LTC TX %1 to %2 / %3 | lat: %4\n", start_sample, end_sample, n_samples, ltc_out_latency.max));
/* all systems go. Now here's the plan:
*
* 1) check if fps has changed
* 2) check direction of encoding, calc speed, re-sample existing buffer
* 3) calculate sample and byte to send aligned to jack-period size
* 4) check if it's the sample/byte that is already in the queue
* 5) if (4) mismatch, re-calculate offset of LTC sample relative to period size
* 6) actual LTC audio output
* 6a) send remaining part of already queued sample; break on n_samples
* 6b) encode new LTC-sample byte
* 6c) goto 6a
* 7) done
*/
// (1) check fps
TimecodeFormat cur_timecode = config.get_timecode_format();
if (cur_timecode != ltc_enc_tcformat) {
DEBUG_TRACE (DEBUG::TXLTC, string_compose("1: TC format mismatch - reinit sr: %1 fps: %2\n", nominal_sample_rate(), timecode_to_frames_per_second(cur_timecode)));
if (ltc_encoder_reinit(ltc_encoder, nominal_sample_rate(),
timecode_to_frames_per_second(cur_timecode),
TV_STANDARD(cur_timecode), 0
)) {
PBD::error << _("LTC encoder: invalid framerate - LTC encoding is disabled for the remainder of this session.") << endmsg;
ltc_tx_cleanup();
return;
}
ltc_encoder_set_filter(ltc_encoder, LTC_RISE_TIME(ltc_speed));
ltc_enc_tcformat = cur_timecode;
ltc_tx_parse_offset();
ltc_tx_reset();
}
/* LTC is max. 30 fps */
if (timecode_to_frames_per_second(cur_timecode) > 30) {
return;
}
// (2) speed & direction
/* speed 0 aka transport stopped is interpreted as rolling forward.
* keep repeating current sample
*/
#define SIGNUM(a) ( (a) < 0 ? -1 : 1)
bool speed_changed = false;
double new_ltc_speed = (end_sample - start_sample) / (double)n_samples;
/* port latency compensation:
* The _generated timecode_ is offset by the port-latency,
* therefore the offset depends on the direction of transport.
*
* latency is compensated by adding it to the timecode to
* be generated. e.g. if the signal will reach the output in
* N samples time from now, generate the timecode for (now + N).
*
* sample-sync is achieved by further calculating the difference
* between the timecode and the session-transport and offsetting the
* buffer.
*
* The timecode is generated directly in the Session process callback
* using _transport_sample (which is the audible frame at the
* output).
*/
samplepos_t cycle_start_sample;
if (new_ltc_speed < 0) {
cycle_start_sample = (start_sample - ltc_out_latency.max);
} else if (new_ltc_speed > 0) {
cycle_start_sample = (start_sample + ltc_out_latency.max);
} else {
/* There is no need to compensate for latency when not rolling
* rather send the accurate NOW timecode
* (LTC encoder compenates latency by sending earlier timecode)
*/
cycle_start_sample = start_sample;
}
/* LTC TV standard offset */
if (new_ltc_speed != 0) {
/* ditto - send "NOW" if not rolling */
cycle_start_sample -= ltc_frame_alignment(samples_per_timecode_frame(), TV_STANDARD(cur_timecode));
}
/* cycle-start may become negative due to latency compensation */
if (cycle_start_sample < 0) { cycle_start_sample = 0; }
if (nominal_sample_rate() != sample_rate()) {
new_ltc_speed *= (double)nominal_sample_rate() / (double)sample_rate();
}
if (SIGNUM(new_ltc_speed) != SIGNUM (ltc_speed)) {
DEBUG_TRACE (DEBUG::TXLTC, "transport changed direction\n");
ltc_tx_reset();
}
if (ltc_speed != new_ltc_speed
/* but only once if, current_speed changes to 0. In that case
* new_ltc_speed is > 0 because (end_sample - start_sample) == jack-period for no-roll
* but ltc_speed will still be 0
*/
//&& (current_speed != 0 || ltc_speed != current_speed)
) {
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: speed change from: %1 to %2\n", ltc_speed, new_ltc_speed));
speed_changed = true;
ltc_encoder_set_filter(ltc_encoder, LTC_RISE_TIME(new_ltc_speed));
}
if (end_sample == start_sample || fabs(new_ltc_speed) < 0.1) {
DEBUG_TRACE (DEBUG::TXLTC, "transport is not rolling or speed < 0.1\n");
/* keep repeating current sample
*
* an LTC generator must be able to continue generating LTC when Ardours transport is in stop
* some machines do odd things if LTC goes away:
* e.g. a tape based machine (video or audio), some think they have gone into park if LTC goes away,
* so unspool the tape from the playhead. That might be inconvenient.
* If LTC keeps arriving they remain in a stop position with the tape on the playhead.
*/
new_ltc_speed = 0;
if (!Config->get_ltc_send_continuously()) {
ltc_speed = new_ltc_speed;
return;
}
if (start_sample != ltc_prev_cycle) {
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: no-roll seek from %1 to %2 (%3)\n", ltc_prev_cycle, start_sample, cycle_start_sample));
ltc_tx_reset();
}
}
if (fabs(new_ltc_speed) > 10.0) {
DEBUG_TRACE (DEBUG::TXLTC, "speed is out of bounds.\n");
ltc_tx_reset();
return;
}
if (ltc_speed == 0 && new_ltc_speed != 0) {
DEBUG_TRACE (DEBUG::TXLTC, "transport started rolling - reset\n");
ltc_tx_reset();
}
/* the timecode duration corresponding to the samples that are still
* in the buffer. Here, the speed of previous cycle is used to calculate
* the alignment at the beginning of this cycle later.
*/
double poff = (ltc_buf_len - ltc_buf_off) * ltc_speed;
if (speed_changed && new_ltc_speed != 0) {
/* we need to re-sample the existing buffer.
* "make space for the en-coder to catch up to the new speed"
*
* since the LTC signal is a rectangular waveform we can simply squeeze it
* by removing samples or duplicating samples /here and there/.
*
* There may be a more elegant way to do this, in fact one could
* simply re-render the buffer using ltc_encoder_encode_byte()
* but that'd require some timecode offset buffer magic,
* which is left for later..
*/
double oldbuflen = (double)(ltc_buf_len - ltc_buf_off);
double newbuflen = (double)(ltc_buf_len - ltc_buf_off) * fabs(ltc_speed / new_ltc_speed);
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: bufOld %1 bufNew %2 | diff %3\n",
(ltc_buf_len - ltc_buf_off), newbuflen, newbuflen - oldbuflen
));
double bufrspdiff = rint(newbuflen - oldbuflen);
if (abs(bufrspdiff) > newbuflen || abs(bufrspdiff) > oldbuflen) {
DEBUG_TRACE (DEBUG::TXLTC, "resampling buffer would destroy information.\n");
ltc_tx_reset();
poff = 0;
} else if (bufrspdiff != 0 && newbuflen > oldbuflen) {
int incnt = 0;
double samples_to_insert = ceil(newbuflen - oldbuflen);
double avg_distance = newbuflen / samples_to_insert;
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: resample buffer insert: %1\n", samples_to_insert));
for (int rp = ltc_buf_off; rp < ltc_buf_len - 1; ++rp) {
const int ro = rp - ltc_buf_off;
if (ro < (incnt*avg_distance)) continue;
const ltcsnd_sample_t v1 = ltc_enc_buf[rp];
const ltcsnd_sample_t v2 = ltc_enc_buf[rp+1];
if (v1 != v2 && ro < ((incnt+1)*avg_distance)) continue;
memmove(&ltc_enc_buf[rp+1], &ltc_enc_buf[rp], ltc_buf_len-rp);
incnt++;
ltc_buf_len++;
}
} else if (bufrspdiff != 0 && newbuflen < oldbuflen) {
double samples_to_remove = ceil(oldbuflen - newbuflen);
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: resample buffer - remove: %1\n", samples_to_remove));
if (oldbuflen <= samples_to_remove) {
ltc_buf_off = ltc_buf_len= 0;
} else {
double avg_distance = newbuflen / samples_to_remove;
int rmcnt = 0;
for (int rp = ltc_buf_off; rp < ltc_buf_len - 1; ++rp) {
const int ro = rp - ltc_buf_off;
if (ro < (rmcnt*avg_distance)) continue;
const ltcsnd_sample_t v1 = ltc_enc_buf[rp];
const ltcsnd_sample_t v2 = ltc_enc_buf[rp+1];
if (v1 != v2 && ro < ((rmcnt+1)*avg_distance)) continue;
memmove(&ltc_enc_buf[rp], &ltc_enc_buf[rp+1], ltc_buf_len-rp-1);
ltc_buf_len--;
rmcnt++;
}
}
}
}
ltc_prev_cycle = start_sample;
ltc_speed = new_ltc_speed;
DEBUG_TRACE (DEBUG::TXLTC, string_compose("2: transport speed %1.\n", ltc_speed));
// (3) bit/sample alignment
Timecode::Time tc_start;
samplepos_t tc_sample_start;
/* calc timecode frame from current position - round down to nearest timecode */
Timecode::sample_to_timecode(cycle_start_sample, tc_start, true, false,
timecode_frames_per_second(),
timecode_drop_frames(),
(double)sample_rate(),
config.get_subframes_per_frame(),
ltc_timecode_negative_offset, ltc_timecode_offset
);
/* convert timecode back to sample-position */
Timecode::timecode_to_sample (tc_start, tc_sample_start, true, false,
(double)sample_rate(),
config.get_subframes_per_frame(),
ltc_timecode_negative_offset, ltc_timecode_offset
);
/* difference between current sample and TC sample in samples */
sampleoffset_t soff = cycle_start_sample - tc_sample_start;
if (new_ltc_speed == 0) {
soff = 0;
}
DEBUG_TRACE (DEBUG::TXLTC, string_compose("3: A3cycle: %1 = A3tc: %2 +off: %3\n",
cycle_start_sample, tc_sample_start, soff));
// (4) check if alignment matches
const double fptcf = samples_per_timecode_frame();
/* maximum difference of bit alignment in audio-samples.
*
* if transport and LTC generator differs more than this, the LTC
* generator will be re-initialized
*
* due to rounding error and variations in LTC-bit duration depending
* on the speed, it can be off by +- ltc_speed audio-samples.
* When the playback speed changes, it can actually reach +- 2 * ltc_speed
* in the cycle _after_ the speed changed. The average delta however is 0.
*/
double maxdiff;
if (transport_master_is_external()) {
maxdiff = transport_master()->resolution();
} else {
maxdiff = ceil(fabs(ltc_speed))*2.0;
if (nominal_sample_rate() != sample_rate()) {
maxdiff *= 3.0;
}
if (ltc_enc_tcformat == Timecode::timecode_23976 || ltc_enc_tcformat == Timecode::timecode_24976) {
maxdiff *= 15.0;
}
}
DEBUG_TRACE (DEBUG::TXLTC, string_compose("4: enc: %1 + %2 - %3 || buf-bytes: %4 enc-byte: %5\n",
ltc_enc_pos, ltc_enc_cnt, poff, (ltc_buf_len - ltc_buf_off), poff, ltc_enc_byte));
DEBUG_TRACE (DEBUG::TXLTC, string_compose("4: enc-pos: %1 | d: %2\n",
ltc_enc_pos + ltc_enc_cnt - poff,
rint(ltc_enc_pos + ltc_enc_cnt - poff) - cycle_start_sample
));
const samplecnt_t wrap24h = 86400. * sample_rate();
if (ltc_enc_pos < 0
|| (ltc_speed != 0 && fabs(fmod(ceil(ltc_enc_pos + ltc_enc_cnt - poff), wrap24h) - (cycle_start_sample % wrap24h)) > maxdiff)
) {
// (5) re-align
ltc_tx_reset();
/* set sample to encode */
SMPTETimecode tc;
tc.hours = tc_start.hours % 24;
tc.mins = tc_start.minutes;
tc.secs = tc_start.seconds;
tc.frame = tc_start.frames;
ltc_encoder_set_timecode(ltc_encoder, &tc);
/* workaround for libltc recognizing 29.97 and 30000/1001 as drop-sample TC.
* In A3 30000/1001 or 30 fps can be drop-sample.
*/
LTCFrame ltcframe;
ltc_encoder_get_frame(ltc_encoder, &ltcframe);
ltcframe.dfbit = timecode_has_drop_frames(cur_timecode)?1:0;
ltc_encoder_set_frame(ltc_encoder, &ltcframe);
DEBUG_TRACE (DEBUG::TXLTC, string_compose("4: now: %1 trs: %2 toff %3\n", cycle_start_sample, tc_sample_start, soff));
int32_t cyc_off;
if (soff < 0 || soff >= fptcf) {
/* session framerate change between (2) and now */
ltc_tx_reset();
return;
}
if (ltc_speed < 0 ) {
/* calculate the byte that starts at or after the current position */
ltc_enc_byte = floor((10.0 * soff) / (fptcf));
ltc_enc_cnt = ltc_enc_byte * fptcf / 10.0;
/* calculate difference between the current position and the byte to send */
cyc_off = soff- ceil(ltc_enc_cnt);
} else {
/* calculate the byte that starts at or after the current position */
ltc_enc_byte = ceil((10.0 * soff) / fptcf);
ltc_enc_cnt = ltc_enc_byte * fptcf / 10.0;
/* calculate difference between the current position and the byte to send */
cyc_off = ceil(ltc_enc_cnt) - soff;
if (ltc_enc_byte == 10) {
ltc_enc_byte = 0;
ltc_encoder_inc_timecode(ltc_encoder);
}
}
DEBUG_TRACE (DEBUG::TXLTC, string_compose("5 restart encoder: soff %1 byte %2 cycoff %3\n",
soff, ltc_enc_byte, cyc_off));
if ( (ltc_speed < 0 && ltc_enc_byte !=9 ) || (ltc_speed >= 0 && ltc_enc_byte !=0 ) ) {
restarting = true;
}
if (cyc_off >= 0 && cyc_off <= (int32_t) n_samples) {
/* offset in this cycle */
txf= rint(cyc_off / fabs(ltc_speed));
memset (out, 0, cyc_off * sizeof(Sample));
} else {
/* resync next cycle */
return;
}
ltc_enc_pos = tc_sample_start % wrap24h;
DEBUG_TRACE (DEBUG::TXLTC, string_compose("5 restart @ %1 + %2 - %3 | byte %4\n",
ltc_enc_pos, ltc_enc_cnt, cyc_off, ltc_enc_byte));
}
else if (ltc_speed != 0 && (fptcf / ltc_speed / 80) > 3 ) {
/* reduce (low freq) jitter.
* The granularity of the LTC encoder speed is 1 byte =
* (samples-per-timecode-sample / 10) audio-samples.
* Thus, tiny speed changes [as produced by some transport masters]
* may not have any effect in the cycle when they occur,
* but they will add up over time.
*
* This is a linear approx to compensate for this jitter
* and prempt re-sync when the drift builds up.
*
* However, for very fast speeds - when 1 LTC bit is
* <= 3 audio-sample - adjusting speed may lead to
* invalid samples.
*
* To do better than this, resampling (or a rewrite of the
* encoder) is required.
*/
ltc_speed -= fmod(((ltc_enc_pos + ltc_enc_cnt - poff) - cycle_start_sample), wrap24h) / engine().sample_rate();
}
// (6) encode and output
while (1) {
#ifdef LTC_GEN_TXDBUG
DEBUG_TRACE (DEBUG::TXLTC, string_compose("6.1 @%1 [ %2 / %3 ]\n", txf, ltc_buf_off, ltc_buf_len));
#endif
// (6a) send remaining buffer
while ((ltc_buf_off < ltc_buf_len) && (txf < n_samples)) {
const float v1 = ltc_enc_buf[ltc_buf_off++] - 128.0;
const Sample val = (Sample) (v1*ltcvol);
out[txf++] = val;
}
#ifdef LTC_GEN_TXDBUG
DEBUG_TRACE (DEBUG::TXLTC, string_compose("6.2 @%1 [ %2 / %3 ]\n", txf, ltc_buf_off, ltc_buf_len));
#endif
if (txf >= n_samples) {
DEBUG_TRACE (DEBUG::TXLTC, string_compose("7 enc: %1 [ %2 / %3 ] byte: %4 spd %5 fpp %6 || nf: %7\n",
ltc_enc_pos, ltc_buf_off, ltc_buf_len, ltc_enc_byte, ltc_speed, n_samples, txf));
break;
}
ltc_buf_len = 0;
ltc_buf_off = 0;
// (6b) encode LTC, bump timecode
if (ltc_speed < 0) {
ltc_enc_byte = (ltc_enc_byte + 9)%10;
if (ltc_enc_byte == 9) {
ltc_encoder_dec_timecode(ltc_encoder);
ltc_tx_recalculate_position();
ltc_enc_cnt = fptcf;
}
}
int enc_samples;
if (restarting) {
/* write zero bytes -- don't touch encoder until we're at a sample-boundary
* otherwise the biphase polarity may be inverted.
*/
enc_samples = fptcf / 10.0;
memset(&ltc_enc_buf[ltc_buf_len], 127, enc_samples * sizeof(ltcsnd_sample_t));
} else {
if (ltc_encoder_encode_byte(ltc_encoder, ltc_enc_byte, (ltc_speed==0)?1.0:(1.0/ltc_speed))) {
DEBUG_TRACE (DEBUG::TXLTC, string_compose("6.3 encoder error byte %1\n", ltc_enc_byte));
ltc_encoder_buffer_flush(ltc_encoder);
ltc_tx_reset();
return;
}
enc_samples = ltc_encoder_get_buffer(ltc_encoder, &(ltc_enc_buf[ltc_buf_len]));
}
#ifdef LTC_GEN_FRAMEDBUG
DEBUG_TRACE (DEBUG::TXLTC, string_compose("6.3 encoded %1 bytes for LTC-byte %2 at spd %3\n", enc_samples, ltc_enc_byte, ltc_speed));
#endif
if (enc_samples <=0) {
DEBUG_TRACE (DEBUG::TXLTC, "6.3 encoder empty buffer.\n");
ltc_encoder_buffer_flush(ltc_encoder);
ltc_tx_reset();
return;
}
ltc_buf_len += enc_samples;
if (ltc_speed < 0)
ltc_enc_cnt -= fptcf/10.0;
else
ltc_enc_cnt += fptcf/10.0;
if (ltc_speed >= 0) {
ltc_enc_byte = (ltc_enc_byte + 1)%10;
if (ltc_enc_byte == 0 && ltc_speed != 0) {
ltc_encoder_inc_timecode(ltc_encoder);
#if 0 /* force fixed parity -- scope debug */
LTCFrame f;
ltc_encoder_get_frame(ltc_encoder, &f);
f.biphase_mark_phase_correction=0;
ltc_encoder_set_frame(ltc_encoder, &f);
#endif
ltc_tx_recalculate_position();
ltc_enc_cnt = 0;
} else if (ltc_enc_byte == 0) {
ltc_enc_cnt = 0;
restarting=false;
}
}
#ifdef LTC_GEN_FRAMEDBUG
DEBUG_TRACE (DEBUG::TXLTC, string_compose("6.4 enc-pos: %1 + %2 [ %4 / %5 ] spd %6\n", ltc_enc_pos, ltc_enc_cnt, ltc_buf_off, ltc_buf_len, ltc_speed));
#endif
}
}