// Some audio formats have a magic cookie associated with them which is required to decompress audio data
// When converting audio, a magic cookie may be returned by the Audio Converter so that it may be stored along with
// the output data -- This is done so that it may then be passed back to the Audio Converter at a later time as required
static void WriteCookie(AudioConverterRef converter, AudioFileID destinationFileID)
{
// grab the cookie from the converter and write it to the destinateion file
UInt32 cookieSize = 0;
OSStatus error = AudioConverterGetPropertyInfo(converter, kAudioConverterCompressionMagicCookie, &cookieSize, NULL);
// if there is an error here, then the format doesn't have a cookie - this is perfectly fine as some formats do not
if (noErr == error && 0 != cookieSize) {
char* cookie = new char [cookieSize];
error = AudioConverterGetProperty(converter, kAudioConverterCompressionMagicCookie, &cookieSize, cookie);
if (noErr == error) {
error = AudioFileSetProperty(destinationFileID, kAudioFilePropertyMagicCookieData, cookieSize, cookie);
if (noErr == error) {
printf("Writing magic cookie to destination file: %ld\n", cookieSize);
} else {
printf("Even though some formats have cookies, some files don't take them and that's OK\n");
}
} else {
printf("Could not Get kAudioConverterCompressionMagicCookie from Audio Converter!\n");
}
delete [] cookie;
}
}
static int ffat_update_ctx(AVCodecContext *avctx)
{
ATDecodeContext *at = avctx->priv_data;
AudioStreamBasicDescription format;
UInt32 size = sizeof(format);
if (!AudioConverterGetProperty(at->converter,
kAudioConverterCurrentInputStreamDescription,
&size, &format)) {
if (format.mSampleRate)
avctx->sample_rate = format.mSampleRate;
avctx->channels = format.mChannelsPerFrame;
avctx->channel_layout = av_get_default_channel_layout(avctx->channels);
avctx->frame_size = format.mFramesPerPacket;
}
if (!AudioConverterGetProperty(at->converter,
kAudioConverterCurrentOutputStreamDescription,
&size, &format)) {
format.mSampleRate = avctx->sample_rate;
format.mChannelsPerFrame = avctx->channels;
AudioConverterSetProperty(at->converter,
kAudioConverterCurrentOutputStreamDescription,
size, &format);
}
if (!AudioConverterGetPropertyInfo(at->converter, kAudioConverterOutputChannelLayout,
&size, NULL) && size) {
AudioChannelLayout *layout = av_malloc(size);
uint64_t layout_mask = 0;
int i;
if (!layout)
return AVERROR(ENOMEM);
AudioConverterGetProperty(at->converter, kAudioConverterOutputChannelLayout,
&size, layout);
if (!(layout = ffat_convert_layout(layout, &size)))
return AVERROR(ENOMEM);
for (i = 0; i < layout->mNumberChannelDescriptions; i++) {
int id = ffat_get_channel_id(layout->mChannelDescriptions[i].mChannelLabel);
if (id < 0)
goto done;
if (layout_mask & (1 << id))
goto done;
layout_mask |= 1 << id;
layout->mChannelDescriptions[i].mChannelFlags = i; // Abusing flags as index
}
avctx->channel_layout = layout_mask;
qsort(layout->mChannelDescriptions, layout->mNumberChannelDescriptions,
sizeof(AudioChannelDescription), &ffat_compare_channel_descriptions);
for (i = 0; i < layout->mNumberChannelDescriptions; i++)
at->channel_map[i] = layout->mChannelDescriptions[i].mChannelFlags;
done:
av_free(layout);
}
if (!avctx->frame_size)
avctx->frame_size = 2048;
return 0;
}
std::vector<uint8_t> AudioConverterX::getCompressionMagicCookie()
{
UInt32 size;
Boolean writable;
CHECKCA(AudioConverterGetPropertyInfo(m_converter.get(),
kAudioConverterCompressionMagicCookie, &size, &writable));
std::vector<uint8_t> vec(size / sizeof(uint8_t));
CHECKCA(AudioConverterGetProperty(m_converter.get(),
kAudioConverterCompressionMagicCookie, &size, vec.data()));
return vec;
}
std::vector<AudioValueRange> AudioConverterX::getApplicableEncodeBitRates()
{
UInt32 size;
Boolean writable;
CHECKCA(AudioConverterGetPropertyInfo(m_converter.get(),
kAudioConverterApplicableEncodeBitRates, &size, &writable));
std::vector<AudioValueRange> vec(size / sizeof(AudioValueRange));
CHECKCA(AudioConverterGetProperty(m_converter.get(),
kAudioConverterApplicableEncodeBitRates, &size, vec.data()));
return vec;
}
std::shared_ptr<AudioChannelLayout> AudioConverterX::getOutputChannelLayout()
{
UInt32 size;
Boolean writable;
CHECKCA(AudioConverterGetPropertyInfo(m_converter.get(),
kAudioConverterOutputChannelLayout, &size, &writable));
std::shared_ptr<AudioChannelLayout> acl(
static_cast<AudioChannelLayout*>(std::malloc(size)),
std::free);
CHECKCA(AudioConverterGetProperty(m_converter.get(),
kAudioConverterOutputChannelLayout, &size, acl.get()));
return acl;
}
// Write output channel layout to destination file
static void WriteDestinationChannelLayout(AudioConverterRef converter, AudioFileID sourceFileID, AudioFileID destinationFileID)
{
UInt32 layoutSize = 0;
bool layoutFromConverter = true;
OSStatus error = AudioConverterGetPropertyInfo(converter, kAudioConverterOutputChannelLayout, &layoutSize, NULL);
// if the Audio Converter doesn't have a layout see if the input file does
if (error || 0 == layoutSize) {
error = AudioFileGetPropertyInfo(sourceFileID, kAudioFilePropertyChannelLayout, &layoutSize, NULL);
layoutFromConverter = false;
}
if (noErr == error && 0 != layoutSize) {
char* layout = new char[layoutSize];
if (layoutFromConverter) {
error = AudioConverterGetProperty(converter, kAudioConverterOutputChannelLayout, &layoutSize, layout);
if (error) printf("Could not Get kAudioConverterOutputChannelLayout from Audio Converter!\n");
} else {
error = AudioFileGetProperty(sourceFileID, kAudioFilePropertyChannelLayout, &layoutSize, layout);
if (error) printf("Could not Get kAudioFilePropertyChannelLayout from source file!\n");
}
if (noErr == error) {
error = AudioFileSetProperty(destinationFileID, kAudioFilePropertyChannelLayout, layoutSize, layout);
if (noErr == error) {
printf("Writing channel layout to destination file: %ld\n", layoutSize);
} else {
printf("Even though some formats have layouts, some files don't take them and that's OK\n");
}
}
delete [] layout;
}
}
JNIEXPORT jint JNICALL Java_com_apple_audio_toolbox_AudioConverter_AudioConverterGetPropertyInfo
(JNIEnv *, jclass, jint inAudioConverter, jint inPropertyID, jint outSize, jint outWritable)
{
return (jint)AudioConverterGetPropertyInfo((AudioConverterRef)inAudioConverter, (AudioConverterPropertyID)inPropertyID, (UInt32 *)outSize, (Boolean *)outWritable);
}
static GVBool gviHardwareInitPlayback(GVIDevice * device)
{
GVIHardwareData * data = (GVIHardwareData *)device->m_data;
UInt32 size;
OSStatus result;
UInt32 primeMethod;
SInt32 channelMap[100];
int i;
// create the array of sources
data->m_playbackSources = gviNewSourceList();
if(!data->m_playbackSources)
return GVFalse;
// get the playback format
size = sizeof(AudioStreamBasicDescription);
result = AudioDeviceGetProperty(device->m_deviceID, 0, false, kAudioDevicePropertyStreamFormat, &size, &data->m_playbackStreamDescriptor);
if(result != noErr)
{
gviFreeSourceList(data->m_playbackSources);
return GVFalse;
}
// create a converter from the GV format to the playback format
result = AudioConverterNew(&GVIVoiceFormat, &data->m_playbackStreamDescriptor, &data->m_playbackConverter);
if(result != noErr)
{
gviFreeSourceList(data->m_playbackSources);
return GVFalse;
}
// set it to do no priming
primeMethod = kConverterPrimeMethod_None;
result = AudioConverterSetProperty(data->m_playbackConverter, kAudioConverterPrimeMethod, sizeof(UInt32), &primeMethod);
if(result != noErr)
{
AudioConverterDispose(data->m_playbackConverter);
gviFreeSourceList(data->m_playbackSources);
return GVFalse;
}
// setup the converter to map the input channel to all output channels
result = AudioConverterGetPropertyInfo(data->m_playbackConverter, kAudioConverterChannelMap, &size, NULL);
if(result == noErr)
{
result = AudioConverterGetProperty(data->m_playbackConverter, kAudioConverterChannelMap, &size, channelMap);
if(result == noErr)
{
for(i = 0 ; i < (size / sizeof(SInt32)) ; i++)
channelMap[i] = 0;
AudioConverterSetProperty(data->m_playbackConverter, kAudioConverterChannelMap, size, channelMap);
}
}
// allocate the playback buffer
data->m_playbackBuffer = (GVSample *)gsimalloc(GVIBytesPerFrame);
if(!data->m_playbackBuffer)
{
AudioConverterDispose(data->m_playbackConverter);
gviFreeSourceList(data->m_playbackSources);
return GVFalse;
}
// add property listener
AudioDeviceAddPropertyListener(device->m_deviceID, 0, false, kAudioDevicePropertyDeviceIsAlive, gviPropertyListener, device);
#if GVI_VOLUME_IN_SOFTWARE
// init volume
data->m_playbackVolume = (GVScalar)1.0;
#endif
return GVTrue;
}
/***********************************************************************
* hb_work_encCoreAudio_init
***********************************************************************
*
**********************************************************************/
int encCoreAudioInit(hb_work_object_t *w, hb_job_t *job, enum AAC_MODE mode)
{
hb_work_private_t *pv = calloc(1, sizeof(hb_work_private_t));
hb_audio_t *audio = w->audio;
AudioStreamBasicDescription input, output;
UInt32 tmp, tmpsiz = sizeof(tmp);
OSStatus err;
w->private_data = pv;
pv->job = job;
// pass the number of channels used into the private work data
pv->nchannels =
hb_mixdown_get_discrete_channel_count(audio->config.out.mixdown);
bzero(&input, sizeof(AudioStreamBasicDescription));
input.mSampleRate = (Float64)audio->config.out.samplerate;
input.mFormatID = kAudioFormatLinearPCM;
input.mFormatFlags = (kLinearPCMFormatFlagIsFloat|kAudioFormatFlagsNativeEndian);
input.mBytesPerPacket = 4 * pv->nchannels;
input.mFramesPerPacket = 1;
input.mBytesPerFrame = input.mBytesPerPacket * input.mFramesPerPacket;
input.mChannelsPerFrame = pv->nchannels;
input.mBitsPerChannel = 32;
bzero(&output, sizeof(AudioStreamBasicDescription));
switch (mode)
{
case AAC_MODE_HE:
output.mFormatID = kAudioFormatMPEG4AAC_HE;
break;
case AAC_MODE_LC:
default:
output.mFormatID = kAudioFormatMPEG4AAC;
break;
}
output.mSampleRate = (Float64)audio->config.out.samplerate;
output.mChannelsPerFrame = pv->nchannels;
// let CoreAudio decide the rest
// initialise encoder
err = AudioConverterNew(&input, &output, &pv->converter);
if (err != noErr)
{
// Retry without the samplerate
bzero(&output, sizeof(AudioStreamBasicDescription));
switch (mode)
{
case AAC_MODE_HE:
output.mFormatID = kAudioFormatMPEG4AAC_HE;
break;
case AAC_MODE_LC:
default:
output.mFormatID = kAudioFormatMPEG4AAC;
break;
}
output.mChannelsPerFrame = pv->nchannels;
err = AudioConverterNew(&input, &output, &pv->converter);
if (err != noErr)
{
hb_log("Error creating an AudioConverter err=%"PRId64" output.mBytesPerFrame=%"PRIu64"",
(int64_t)err, (uint64_t)output.mBytesPerFrame);
*job->done_error = HB_ERROR_UNKNOWN;
*job->die = 1;
return -1;
}
}
// set encoder quality to maximum
tmp = kAudioConverterQuality_Max;
AudioConverterSetProperty(pv->converter, kAudioConverterCodecQuality,
sizeof(tmp), &tmp);
if (audio->config.out.bitrate > 0)
{
// set encoder bitrate control mode to constrained variable
tmp = kAudioCodecBitRateControlMode_VariableConstrained;
AudioConverterSetProperty(pv->converter,
kAudioCodecPropertyBitRateControlMode,
sizeof(tmp), &tmp);
// get available bitrates
AudioValueRange *bitrates;
ssize_t bitrateCounts;
err = AudioConverterGetPropertyInfo(pv->converter,
kAudioConverterApplicableEncodeBitRates,
&tmpsiz, NULL);
bitrates = malloc(tmpsiz);
err = AudioConverterGetProperty(pv->converter,
kAudioConverterApplicableEncodeBitRates,
&tmpsiz, bitrates);
bitrateCounts = tmpsiz / sizeof(AudioValueRange);
// set bitrate
tmp = audio->config.out.bitrate * 1000;
if (tmp < bitrates[0].mMinimum)
tmp = bitrates[0].mMinimum;
if (tmp > bitrates[bitrateCounts-1].mMinimum)
tmp = bitrates[bitrateCounts-1].mMinimum;
free(bitrates);
if (tmp != audio->config.out.bitrate * 1000)
{
hb_log("encCoreAudioInit: sanitizing track %d audio bitrate %d to %"PRIu32"",
audio->config.out.track, audio->config.out.bitrate, tmp / 1000);
}
AudioConverterSetProperty(pv->converter,
kAudioConverterEncodeBitRate,
sizeof(tmp), &tmp);
}
else if (audio->config.out.quality >= 0)
{
if (mode != AAC_MODE_LC)
{
hb_error("encCoreAudioInit: internal error, VBR set but not applicable");
return 1;
}
// set encoder bitrate control mode to variable
tmp = kAudioCodecBitRateControlMode_Variable;
AudioConverterSetProperty(pv->converter,
kAudioCodecPropertyBitRateControlMode,
sizeof(tmp), &tmp);
// set quality
tmp = audio->config.out.quality;
AudioConverterSetProperty(pv->converter,
kAudioCodecPropertySoundQualityForVBR,
sizeof(tmp), &tmp);
}
else
{
hb_error("encCoreAudioInit: internal error, bitrate/quality not set");
return 1;
}
// get real input
tmpsiz = sizeof(input);
AudioConverterGetProperty(pv->converter,
kAudioConverterCurrentInputStreamDescription,
&tmpsiz, &input);
// get real output
tmpsiz = sizeof(output);
AudioConverterGetProperty(pv->converter,
kAudioConverterCurrentOutputStreamDescription,
&tmpsiz, &output);
// set sizes
pv->isamplesiz = input.mBytesPerPacket;
pv->isamples = output.mFramesPerPacket;
pv->osamplerate = output.mSampleRate;
audio->config.out.samples_per_frame = pv->isamples;
// channel remapping
pv->remap = hb_audio_remap_init(AV_SAMPLE_FMT_FLT, &hb_aac_chan_map,
audio->config.in.channel_map);
if (pv->remap == NULL)
{
hb_error("encCoreAudioInit: hb_audio_remap_init() failed");
}
uint64_t layout = hb_ff_mixdown_xlat(audio->config.out.mixdown, NULL);
hb_audio_remap_set_channel_layout(pv->remap, layout);
// get maximum output size
AudioConverterGetProperty(pv->converter,
kAudioConverterPropertyMaximumOutputPacketSize,
&tmpsiz, &tmp);
pv->omaxpacket = tmp;
// get magic cookie (elementary stream descriptor)
tmp = HB_CONFIG_MAX_SIZE;
AudioConverterGetProperty(pv->converter,
kAudioConverterCompressionMagicCookie,
&tmp, w->config->extradata.bytes);
// CoreAudio returns a complete ESDS, but we only need
// the DecoderSpecific info.
UInt8* buffer = NULL;
ReadESDSDescExt(w->config->extradata.bytes, &buffer, &tmpsiz, 0);
w->config->extradata.length = tmpsiz;
memmove(w->config->extradata.bytes, buffer, w->config->extradata.length);
free(buffer);
pv->list = hb_list_init();
pv->buf = NULL;
return 0;
}
/***********************************************************************
* hb_work_encCoreAudio_init
***********************************************************************
*
**********************************************************************/
int encCoreAudioInit( hb_work_object_t * w, hb_job_t * job, enum AAC_MODE mode )
{
hb_work_private_t * pv = calloc( 1, sizeof( hb_work_private_t ) );
hb_audio_t * audio = w->audio;
AudioStreamBasicDescription input, output;
UInt32 tmp, tmpsiz = sizeof( tmp );
OSStatus err;
w->private_data = pv;
pv->job = job;
// pass the number of channels used into the private work data
pv->nchannels = HB_AMIXDOWN_GET_DISCRETE_CHANNEL_COUNT( audio->config.out.mixdown );
bzero( &input, sizeof( AudioStreamBasicDescription ) );
input.mSampleRate = ( Float64 ) audio->config.out.samplerate;
input.mFormatID = kAudioFormatLinearPCM;
input.mFormatFlags = kLinearPCMFormatFlagIsFloat | kAudioFormatFlagsNativeEndian;
input.mBytesPerPacket = 4 * pv->nchannels;
input.mFramesPerPacket = 1;
input.mBytesPerFrame = input.mBytesPerPacket * input.mFramesPerPacket;
input.mChannelsPerFrame = pv->nchannels;
input.mBitsPerChannel = 32;
bzero( &output, sizeof( AudioStreamBasicDescription ) );
switch ( mode )
{
case AAC_MODE_HE:
output.mFormatID = kAudioFormatMPEG4AAC_HE;
break;
case AAC_MODE_LC:
default:
output.mFormatID = kAudioFormatMPEG4AAC;
break;
}
output.mSampleRate = ( Float64 ) audio->config.out.samplerate;
output.mChannelsPerFrame = pv->nchannels;
// let CoreAudio decide the rest...
// initialise encoder
err = AudioConverterNew( &input, &output, &pv->converter );
if( err != noErr)
{
// Retry without the samplerate
bzero( &output, sizeof( AudioStreamBasicDescription ) );
switch ( mode )
{
case AAC_MODE_HE:
output.mFormatID = kAudioFormatMPEG4AAC_HE;
break;
case AAC_MODE_LC:
default:
output.mFormatID = kAudioFormatMPEG4AAC;
break;
}
output.mChannelsPerFrame = pv->nchannels;
err = AudioConverterNew( &input, &output, &pv->converter );
if( err != noErr)
{
hb_log( "Error creating an AudioConverter err=%"PRId64" %"PRIu64, (int64_t)err, (uint64_t)output.mBytesPerFrame );
*job->die = 1;
return 0;
}
}
if( ( audio->config.out.mixdown == HB_AMIXDOWN_6CH ) && ( audio->config.in.codec == HB_ACODEC_AC3) )
{
SInt32 channelMap[6] = { 2, 1, 3, 4, 5, 0 };
AudioConverterSetProperty( pv->converter, kAudioConverterChannelMap,
sizeof( channelMap ), channelMap );
}
// set encoder quality to maximum
tmp = kAudioConverterQuality_Max;
AudioConverterSetProperty( pv->converter, kAudioConverterCodecQuality,
sizeof( tmp ), &tmp );
// set encoder bitrate control mode to constrained variable
tmp = kAudioCodecBitRateControlMode_VariableConstrained;
AudioConverterSetProperty( pv->converter, kAudioCodecPropertyBitRateControlMode,
sizeof( tmp ), &tmp );
// get available bitrates
AudioValueRange *bitrates;
ssize_t bitrateCounts;
err = AudioConverterGetPropertyInfo( pv->converter, kAudioConverterApplicableEncodeBitRates,
&tmpsiz, NULL);
bitrates = malloc( tmpsiz );
err = AudioConverterGetProperty( pv->converter, kAudioConverterApplicableEncodeBitRates,
&tmpsiz, bitrates);
bitrateCounts = tmpsiz / sizeof( AudioValueRange );
// set bitrate
tmp = audio->config.out.bitrate * 1000;
if( tmp < bitrates[0].mMinimum )
tmp = bitrates[0].mMinimum;
if( tmp > bitrates[bitrateCounts-1].mMinimum )
tmp = bitrates[bitrateCounts-1].mMinimum;
free( bitrates );
if( tmp != audio->config.out.bitrate * 1000 )
hb_log( "encca_aac: sanitizing track %d audio bitrate %d to %"PRIu32"",
audio->config.out.track, audio->config.out.bitrate, tmp/1000 );
AudioConverterSetProperty( pv->converter, kAudioConverterEncodeBitRate,
sizeof( tmp ), &tmp );
// get real input
tmpsiz = sizeof( input );
AudioConverterGetProperty( pv->converter,
kAudioConverterCurrentInputStreamDescription,
&tmpsiz, &input );
// get real output
tmpsiz = sizeof( output );
AudioConverterGetProperty( pv->converter,
kAudioConverterCurrentOutputStreamDescription,
&tmpsiz, &output );
// set sizes
pv->isamplesiz = input.mBytesPerPacket;
pv->isamples = output.mFramesPerPacket;
pv->osamplerate = output.mSampleRate;
// get maximum output size
AudioConverterGetProperty( pv->converter,
kAudioConverterPropertyMaximumOutputPacketSize,
&tmpsiz, &tmp );
pv->omaxpacket = tmp;
// get magic cookie (elementary stream descriptor)
tmp = HB_CONFIG_MAX_SIZE;
AudioConverterGetProperty( pv->converter,
kAudioConverterCompressionMagicCookie,
&tmp, w->config->aac.bytes );
// CoreAudio returns a complete ESDS, but we only need
// the DecoderSpecific info.
UInt8* buffer = NULL;
ReadESDSDescExt(w->config->aac.bytes, &buffer, &tmpsiz, 0);
w->config->aac.length = tmpsiz;
memmove( w->config->aac.bytes, buffer,
w->config->aac.length );
pv->list = hb_list_init();
pv->buf = NULL;
return 0;
}
static av_cold int ffat_init_encoder(AVCodecContext *avctx)
{
ATDecodeContext *at = avctx->priv_data;
OSStatus status;
AudioStreamBasicDescription in_format = {
.mSampleRate = avctx->sample_rate,
.mFormatID = kAudioFormatLinearPCM,
.mFormatFlags = ((avctx->sample_fmt == AV_SAMPLE_FMT_FLT ||
avctx->sample_fmt == AV_SAMPLE_FMT_DBL) ? kAudioFormatFlagIsFloat
: avctx->sample_fmt == AV_SAMPLE_FMT_U8 ? 0
: kAudioFormatFlagIsSignedInteger)
| kAudioFormatFlagIsPacked,
.mBytesPerPacket = av_get_bytes_per_sample(avctx->sample_fmt) * avctx->channels,
.mFramesPerPacket = 1,
.mBytesPerFrame = av_get_bytes_per_sample(avctx->sample_fmt) * avctx->channels,
.mChannelsPerFrame = avctx->channels,
.mBitsPerChannel = av_get_bytes_per_sample(avctx->sample_fmt) * 8,
};
AudioStreamBasicDescription out_format = {
.mSampleRate = avctx->sample_rate,
.mFormatID = ffat_get_format_id(avctx->codec_id, avctx->profile),
.mChannelsPerFrame = in_format.mChannelsPerFrame,
};
UInt32 layout_size = sizeof(AudioChannelLayout) +
sizeof(AudioChannelDescription) * avctx->channels;
AudioChannelLayout *channel_layout = av_malloc(layout_size);
if (!channel_layout)
return AVERROR(ENOMEM);
if (avctx->codec_id == AV_CODEC_ID_ILBC) {
int mode = get_ilbc_mode(avctx);
out_format.mFramesPerPacket = 8000 * mode / 1000;
out_format.mBytesPerPacket = (mode == 20 ? 38 : 50);
}
status = AudioConverterNew(&in_format, &out_format, &at->converter);
if (status != 0) {
av_log(avctx, AV_LOG_ERROR, "AudioToolbox init error: %i\n", (int)status);
av_free(channel_layout);
return AVERROR_UNKNOWN;
}
if (!avctx->channel_layout)
avctx->channel_layout = av_get_default_channel_layout(avctx->channels);
if ((status = remap_layout(channel_layout, avctx->channel_layout, avctx->channels)) < 0) {
av_log(avctx, AV_LOG_ERROR, "Invalid channel layout\n");
av_free(channel_layout);
return status;
}
if (AudioConverterSetProperty(at->converter, kAudioConverterInputChannelLayout,
layout_size, channel_layout)) {
av_log(avctx, AV_LOG_ERROR, "Unsupported input channel layout\n");
av_free(channel_layout);
return AVERROR(EINVAL);
}
if (avctx->codec_id == AV_CODEC_ID_AAC) {
int tag = get_aac_tag(avctx->channel_layout);
if (tag) {
channel_layout->mChannelLayoutTag = tag;
channel_layout->mNumberChannelDescriptions = 0;
}
}
if (AudioConverterSetProperty(at->converter, kAudioConverterOutputChannelLayout,
layout_size, channel_layout)) {
av_log(avctx, AV_LOG_ERROR, "Unsupported output channel layout\n");
av_free(channel_layout);
return AVERROR(EINVAL);
}
av_free(channel_layout);
if (avctx->bits_per_raw_sample)
AudioConverterSetProperty(at->converter,
kAudioConverterPropertyBitDepthHint,
sizeof(avctx->bits_per_raw_sample),
&avctx->bits_per_raw_sample);
#if !TARGET_OS_IPHONE
if (at->mode == -1)
at->mode = (avctx->flags & AV_CODEC_FLAG_QSCALE) ?
kAudioCodecBitRateControlMode_Variable :
kAudioCodecBitRateControlMode_Constant;
AudioConverterSetProperty(at->converter, kAudioCodecPropertyBitRateControlMode,
sizeof(at->mode), &at->mode);
if (at->mode == kAudioCodecBitRateControlMode_Variable) {
int q = avctx->global_quality / FF_QP2LAMBDA;
if (q < 0 || q > 14) {
av_log(avctx, AV_LOG_WARNING,
"VBR quality %d out of range, should be 0-14\n", q);
q = av_clip(q, 0, 14);
}
q = 127 - q * 9;
AudioConverterSetProperty(at->converter, kAudioCodecPropertySoundQualityForVBR,
sizeof(q), &q);
} else
#endif
if (avctx->bit_rate > 0) {
UInt32 rate = avctx->bit_rate;
UInt32 size;
status = AudioConverterGetPropertyInfo(at->converter,
kAudioConverterApplicableEncodeBitRates,
&size, NULL);
if (!status && size) {
UInt32 new_rate = rate;
int count;
int i;
AudioValueRange *ranges = av_malloc(size);
if (!ranges)
return AVERROR(ENOMEM);
AudioConverterGetProperty(at->converter,
kAudioConverterApplicableEncodeBitRates,
&size, ranges);
count = size / sizeof(AudioValueRange);
for (i = 0; i < count; i++) {
AudioValueRange *range = &ranges[i];
if (rate >= range->mMinimum && rate <= range->mMaximum) {
new_rate = rate;
break;
} else if (rate > range->mMaximum) {
new_rate = range->mMaximum;
} else {
new_rate = range->mMinimum;
break;
}
}
if (new_rate != rate) {
av_log(avctx, AV_LOG_WARNING,
"Bitrate %u not allowed; changing to %u\n", rate, new_rate);
rate = new_rate;
}
av_free(ranges);
}
AudioConverterSetProperty(at->converter, kAudioConverterEncodeBitRate,
sizeof(rate), &rate);
}
at->quality = 96 - at->quality * 32;
AudioConverterSetProperty(at->converter, kAudioConverterCodecQuality,
sizeof(at->quality), &at->quality);
if (!AudioConverterGetPropertyInfo(at->converter, kAudioConverterCompressionMagicCookie,
&avctx->extradata_size, NULL) &&
avctx->extradata_size) {
int extradata_size = avctx->extradata_size;
uint8_t *extradata;
if (!(avctx->extradata = av_mallocz(avctx->extradata_size + AV_INPUT_BUFFER_PADDING_SIZE)))
return AVERROR(ENOMEM);
if (avctx->codec_id == AV_CODEC_ID_ALAC) {
avctx->extradata_size = 0x24;
AV_WB32(avctx->extradata, 0x24);
AV_WB32(avctx->extradata + 4, MKBETAG('a','l','a','c'));
extradata = avctx->extradata + 12;
avctx->extradata_size = 0x24;
} else {
extradata = avctx->extradata;
}
status = AudioConverterGetProperty(at->converter,
kAudioConverterCompressionMagicCookie,
&extradata_size, extradata);
if (status != 0) {
av_log(avctx, AV_LOG_ERROR, "AudioToolbox cookie error: %i\n", (int)status);
return AVERROR_UNKNOWN;
} else if (avctx->codec_id == AV_CODEC_ID_AAC) {
GetByteContext gb;
int tag, len;
bytestream2_init(&gb, extradata, extradata_size);
do {
len = read_descr(&gb, &tag);
if (tag == MP4DecConfigDescrTag) {
bytestream2_skip(&gb, 13);
len = read_descr(&gb, &tag);
if (tag == MP4DecSpecificDescrTag) {
len = FFMIN(gb.buffer_end - gb.buffer, len);
memmove(extradata, gb.buffer, len);
avctx->extradata_size = len;
break;
}
} else if (tag == MP4ESDescrTag) {
int flags;
bytestream2_skip(&gb, 2);
flags = bytestream2_get_byte(&gb);
if (flags & 0x80) //streamDependenceFlag
bytestream2_skip(&gb, 2);
if (flags & 0x40) //URL_Flag
bytestream2_skip(&gb, bytestream2_get_byte(&gb));
if (flags & 0x20) //OCRstreamFlag
bytestream2_skip(&gb, 2);
}
} while (bytestream2_get_bytes_left(&gb));
} else if (avctx->codec_id != AV_CODEC_ID_ALAC) {
avctx->extradata_size = extradata_size;
}
}
ffat_update_ctx(avctx);
#if !TARGET_OS_IPHONE && defined(__MAC_10_9)
if (at->mode == kAudioCodecBitRateControlMode_Variable && avctx->rc_max_rate) {
UInt32 max_size = avctx->rc_max_rate * avctx->frame_size / avctx->sample_rate;
if (max_size)
AudioConverterSetProperty(at->converter, kAudioCodecPropertyPacketSizeLimitForVBR,
sizeof(max_size), &max_size);
}
#endif
ff_af_queue_init(avctx, &at->afq);
return 0;
}
static OSStatus ffat_encode_callback(AudioConverterRef converter, UInt32 *nb_packets,
AudioBufferList *data,
AudioStreamPacketDescription **packets,
void *inctx)
{
AVCodecContext *avctx = inctx;
ATDecodeContext *at = avctx->priv_data;
AVFrame *frame;
if (!at->frame_queue.available) {
if (at->eof) {
*nb_packets = 0;
return 0;
} else {
*nb_packets = 0;
return 1;
}
}
frame = ff_bufqueue_get(&at->frame_queue);
data->mNumberBuffers = 1;
data->mBuffers[0].mNumberChannels = avctx->channels;
data->mBuffers[0].mDataByteSize = frame->nb_samples *
av_get_bytes_per_sample(avctx->sample_fmt) *
avctx->channels;
data->mBuffers[0].mData = frame->data[0];
if (*nb_packets > frame->nb_samples)
*nb_packets = frame->nb_samples;
ff_bufqueue_add(avctx, &at->used_frame_queue, frame);
return 0;
}
static int ffat_encode(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
ATDecodeContext *at = avctx->priv_data;
OSStatus ret;
AudioBufferList out_buffers = {
.mNumberBuffers = 1,
.mBuffers = {
{
.mNumberChannels = avctx->channels,
.mDataByteSize = at->pkt_size,
}
}
};
// _______________________________________________________________________________________
//
// called to create the file -- or update its format/channel layout/properties based on an encoder
// setting change
void CAAudioFile::FileFormatChanged(const FSRef *parentDir, CFStringRef filename, AudioFileTypeID filetype)
{
LOG_FUNCTION("CAAudioFile::FileFormatChanged", "%p", this);
XThrowIf(mMode != kPreparingToCreate && mMode != kPreparingToWrite, kExtAudioFileError_InvalidOperationOrder, "new file not prepared");
UInt32 propertySize;
OSStatus err;
AudioStreamBasicDescription saveFileDataFormat = mFileDataFormat;
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Specified file data format");
#endif
// Find out the actual format the converter will produce. This is necessary in
// case the bitrate has forced a lower sample rate, which needs to be set correctly
// in the stream description passed to AudioFileCreate.
if (mConverter != NULL) {
propertySize = sizeof(AudioStreamBasicDescription);
Float64 origSampleRate = mFileDataFormat.mSampleRate;
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterCurrentOutputStreamDescription, &propertySize, &mFileDataFormat), "get audio converter's output stream description");
// do the same for the channel layout being output by the converter
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Converter output");
#endif
if (fiszero(mFileDataFormat.mSampleRate))
mFileDataFormat.mSampleRate = origSampleRate;
err = AudioConverterGetPropertyInfo(mConverter, kAudioConverterOutputChannelLayout, &propertySize, NULL);
if (err == noErr && propertySize > 0) {
AudioChannelLayout *layout = static_cast<AudioChannelLayout *>(malloc(propertySize));
err = AudioConverterGetProperty(mConverter, kAudioConverterOutputChannelLayout, &propertySize, layout);
if (err) {
free(layout);
XThrow(err, "couldn't get audio converter's output channel layout");
}
mFileChannelLayout = layout;
#if VERBOSE_CHANNELMAP
printf("got new file's channel layout from converter: %s\n", CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
free(layout);
}
}
// create the output file
if (mMode == kPreparingToCreate) {
CAStreamBasicDescription newFileDataFormat = mFileDataFormat;
if (fiszero(newFileDataFormat.mSampleRate))
newFileDataFormat.mSampleRate = 44100; // just make something up for now
#if VERBOSE_CONVERTER
newFileDataFormat.PrintFormat(stdout, "", "Applied to new file");
#endif
XThrowIfError(AudioFileCreate(parentDir, filename, filetype, &newFileDataFormat, 0, &mFSRef, &mAudioFile), "create audio file");
mMode = kPreparingToWrite;
mOwnOpenFile = true;
} else if (saveFileDataFormat != mFileDataFormat || fnotequal(saveFileDataFormat.mSampleRate, mFileDataFormat.mSampleRate)) {
// second check must be explicit since operator== on ASBD treats SR of zero as "don't care"
if (fiszero(mFileDataFormat.mSampleRate))
mFileDataFormat.mSampleRate = mClientDataFormat.mSampleRate;
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Applied to new file");
#endif
XThrowIf(fiszero(mFileDataFormat.mSampleRate), kExtAudioFileError_InvalidDataFormat, "file's sample rate is 0");
XThrowIfError(AudioFileSetProperty(mAudioFile, kAudioFilePropertyDataFormat, sizeof(AudioStreamBasicDescription), &mFileDataFormat), "couldn't update file's data format");
}
UInt32 deferSizeUpdates = 1;
err = AudioFileSetProperty(mAudioFile, kAudioFilePropertyDeferSizeUpdates, sizeof(UInt32), &deferSizeUpdates);
if (mConverter != NULL) {
// encoder
// get the magic cookie, if any, from the converter
delete[] mMagicCookie; mMagicCookie = NULL;
mMagicCookieSize = 0;
err = AudioConverterGetPropertyInfo(mConverter, kAudioConverterCompressionMagicCookie, &propertySize, NULL);
// we can get a noErr result and also a propertySize == 0
// -- if the file format does support magic cookies, but this file doesn't have one.
if (err == noErr && propertySize > 0) {
mMagicCookie = new Byte[propertySize];
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterCompressionMagicCookie, &propertySize, mMagicCookie), "get audio converter's magic cookie");
mMagicCookieSize = propertySize; // the converter lies and tell us the wrong size
// now set the magic cookie on the output file
UInt32 willEatTheCookie = false;
// the converter wants to give us one; will the file take it?
err = AudioFileGetPropertyInfo(mAudioFile, kAudioFilePropertyMagicCookieData,
NULL, &willEatTheCookie);
if (err == noErr && willEatTheCookie) {
#if VERBOSE_CONVERTER
printf("Setting cookie on encoded file\n");
#endif
XThrowIfError(AudioFileSetProperty(mAudioFile, kAudioFilePropertyMagicCookieData, mMagicCookieSize, mMagicCookie), "set audio file's magic cookie");
}
}
// get maximum packet size
propertySize = sizeof(UInt32);
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterPropertyMaximumOutputPacketSize, &propertySize, &mFileMaxPacketSize), "get audio converter's maximum output packet size");
AllocateBuffers(true /* okToFail */);
} else {
InitFileMaxPacketSize();
}
if (mFileChannelLayout.IsValid() && mFileChannelLayout.NumberChannels() > 2) {
// don't bother tagging mono/stereo files
UInt32 isWritable;
err = AudioFileGetPropertyInfo(mAudioFile, kAudioFilePropertyChannelLayout, NULL, &isWritable);
if (!err && isWritable) {
#if VERBOSE_CHANNELMAP
printf("writing file's channel layout: %s\n", CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
err = AudioFileSetProperty(mAudioFile, kAudioFilePropertyChannelLayout,
mFileChannelLayout.Size(), &mFileChannelLayout.Layout());
if (err)
CAXException::Warning("could not set the file's channel layout", err);
} else {
#if VERBOSE_CHANNELMAP
printf("file won't accept a channel layout (write)\n");
#endif
}
}
UpdateClientMaxPacketSize(); // also sets mFrame0Offset
mPacketMark = 0;
mFrameMark = 0;
}
// Will set mChannelLayout if a channel layout could properly be identified
// and is supported.
nsresult
AppleATDecoder::SetupChannelLayout()
{
MOZ_ASSERT(mTaskQueue->IsCurrentThreadIn());
// Determine the channel layout.
UInt32 propertySize;
UInt32 size;
OSStatus status =
AudioConverterGetPropertyInfo(mConverter,
kAudioConverterOutputChannelLayout,
&propertySize, NULL);
if (status || !propertySize) {
LOG("Couldn't get channel layout property (%s)", FourCC2Str(status));
return NS_ERROR_FAILURE;
}
auto data = MakeUnique<uint8_t[]>(propertySize);
size = propertySize;
status =
AudioConverterGetProperty(mConverter, kAudioConverterInputChannelLayout,
&size, data.get());
if (status || size != propertySize) {
LOG("Couldn't get channel layout property (%s)",
FourCC2Str(status));
return NS_ERROR_FAILURE;
}
AudioChannelLayout* layout =
reinterpret_cast<AudioChannelLayout*>(data.get());
AudioChannelLayoutTag tag = layout->mChannelLayoutTag;
// if tag is kAudioChannelLayoutTag_UseChannelDescriptions then the structure
// directly contains the the channel layout mapping.
// If tag is kAudioChannelLayoutTag_UseChannelBitmap then the layout will
// be defined via the bitmap and can be retrieved using
// kAudioFormatProperty_ChannelLayoutForBitmap property.
// Otherwise the tag itself describes the layout.
if (tag != kAudioChannelLayoutTag_UseChannelDescriptions) {
AudioFormatPropertyID property =
tag == kAudioChannelLayoutTag_UseChannelBitmap
? kAudioFormatProperty_ChannelLayoutForBitmap
: kAudioFormatProperty_ChannelLayoutForTag;
if (property == kAudioFormatProperty_ChannelLayoutForBitmap) {
status =
AudioFormatGetPropertyInfo(property,
sizeof(UInt32), &layout->mChannelBitmap,
&propertySize);
} else {
status =
AudioFormatGetPropertyInfo(property,
sizeof(AudioChannelLayoutTag), &tag,
&propertySize);
}
if (status || !propertySize) {
LOG("Couldn't get channel layout property info (%s:%s)",
FourCC2Str(property), FourCC2Str(status));
return NS_ERROR_FAILURE;
}
data = MakeUnique<uint8_t[]>(propertySize);
layout = reinterpret_cast<AudioChannelLayout*>(data.get());
size = propertySize;
if (property == kAudioFormatProperty_ChannelLayoutForBitmap) {
status = AudioFormatGetProperty(property,
sizeof(UInt32), &layout->mChannelBitmap,
&size, layout);
} else {
status = AudioFormatGetProperty(property,
sizeof(AudioChannelLayoutTag), &tag,
&size, layout);
}
if (status || size != propertySize) {
LOG("Couldn't get channel layout property (%s:%s)",
FourCC2Str(property), FourCC2Str(status));
return NS_ERROR_FAILURE;
}
// We have retrieved the channel layout from the tag or bitmap.
// We can now directly use the channel descriptions.
layout->mChannelLayoutTag = kAudioChannelLayoutTag_UseChannelDescriptions;
}
if (layout->mNumberChannelDescriptions > MAX_AUDIO_CHANNELS ||
layout->mNumberChannelDescriptions != mOutputFormat.mChannelsPerFrame) {
LOG("Nonsensical channel layout or not matching the original channel number");
return NS_ERROR_FAILURE;
}
AudioConfig::Channel channels[MAX_AUDIO_CHANNELS];
for (uint32_t i = 0; i < layout->mNumberChannelDescriptions; i++) {
AudioChannelLabel id = layout->mChannelDescriptions[i].mChannelLabel;
AudioConfig::Channel channel = ConvertChannelLabel(id);
channels[i] = channel;
}
mChannelLayout =
MakeUnique<AudioConfig::ChannelLayout>(mOutputFormat.mChannelsPerFrame,
channels);
return NS_OK;
}
