Autres/audacity
1
0
Fork 0
mirror of https://github.com/cookiengineer/audacity synced 2025-07-12 14:47:43 +02:00
Code Issues Releases Wiki Activity

Fix truncated decode of mp3 files on import

Browse Source 
The MAD decoder will not decode the final frame in an mp3 stream unless
it has a small amount of padding afterwards (MAD_BUFFER_GUARD bytes,
actually 8). Without this, it loses sync before returning any decoded
data from the final frame. The result is that the imported audio is
truncated by up to 1152 samples.

This commit addresses that by using a final round of input callback
calls to provide the last MAD_BUFFER_GUARD bytes after the underlying
file has reached eof. The logic is based on madplay.
This commit is contained in:
Chris Cannam 2016-11-26 17:41:07 +00:00
parent 639bfcbad4
commit 30eaa0d52c
1 changed files with 32 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
src/import/ImportMP3.cpp
View File

@ -95,12 +95,14 @@ extern "C" {
struct private_data {
wxFile *file; /* the file containing the mp3 data we're feeding the encoder */
unsigned char *inputBuffer;
int inputBufferFill; /* amount of data in inputBuffer */
TrackFactory *trackFactory;
TrackHolders channels;
ProgressDialog *progress;
unsigned numChannels;
int updateResult;
bool id3checked;
bool eof; /* having supplied both underlying file and guard pad data */
};
class MP3ImportPlugin final : public ImportPlugin
@ -216,11 +218,13 @@ int MP3ImportFileHandle::Import(TrackFactory *trackFactory, TrackHolders &outTra
mPrivateData.file = mFile.get();
mPrivateData.inputBuffer = new unsigned char [INPUT_BUFFER_SIZE];
mPrivateData.inputBufferFill = 0;
mPrivateData.progress = mProgress.get();
mPrivateData.updateResult= eProgressSuccess;
mPrivateData.id3checked = false;
mPrivateData.numChannels = 0;
mPrivateData.trackFactory= trackFactory;
mPrivateData.eof = false;
mad_decoder_init(&mDecoder, &mPrivateData, input_cb, 0, 0, output_cb, error_cb, 0);
@ -404,7 +408,10 @@ enum mad_flow input_cb(void *_data, struct mad_stream *stream)
if(data->updateResult != eProgressSuccess)
return MAD_FLOW_STOP;
if(data->file->Eof()) {
if (data->eof) {
/* different from data->File->Eof(), this means the underlying
file has reached eof *and* we have subsequently supplied the
final padding zeros */
return MAD_FLOW_STOP;
}
@ -431,20 +438,41 @@ enum mad_flow input_cb(void *_data, struct mad_stream *stream)
* mad_stream_buffer()"
* -- Rob Leslie, on the mad-dev mailing list */
unsigned int unconsumedBytes;
if(stream->next_frame) {
unconsumedBytes = data->inputBuffer + INPUT_BUFFER_SIZE - stream->next_frame;
int unconsumedBytes;
if (stream->next_frame) {
/* we must use inputBufferFill instead of INPUT_BUFFER_SIZE here
because the final buffer of the file may be only partially
filled, and we would otherwise be providing too much input
after eof */
unconsumedBytes = data->inputBuffer + data->inputBufferFill
- stream->next_frame;
memmove(data->inputBuffer, stream->next_frame, unconsumedBytes);
}
else
unconsumedBytes = 0;
if (data->file->Eof() &&
(unconsumedBytes + MAD_BUFFER_GUARD < INPUT_BUFFER_SIZE)) {
/* supply the requisite MAD_BUFFER_GUARD zero bytes to ensure
the final frame gets decoded properly, then finish */
memset(data->inputBuffer + unconsumedBytes, 0, MAD_BUFFER_GUARD);
mad_stream_buffer
(stream, data->inputBuffer, MAD_BUFFER_GUARD + unconsumedBytes);
data->eof = true; /* so on next call, we will tell mad to stop */
return MAD_FLOW_CONTINUE;
}
off_t read = data->file->Read(data->inputBuffer + unconsumedBytes,
INPUT_BUFFER_SIZE - unconsumedBytes);
mad_stream_buffer(stream, data->inputBuffer, read + unconsumedBytes);
data->inputBufferFill = int(read + unconsumedBytes);
return MAD_FLOW_CONTINUE;
}