Latency is set via 'sf_command' using 'SFC_SET_OGG_PAGE_LATENCY_MS'.
Max average latency set at 1600ms, min average latency set at 50ms.
Default latency set at 1000ms.
Closes: https://github.com/erikd/libsndfile/pull/512
* Fix Ogg and Opus imported targets
* Better SndFileConfig.cmake module with dependency checks
* Fix config-file package installation directory
* Prefer config-file package mode for Windows CI builds
* Update README.md
* Allow CMAKE_MSVC_RUNTIME_LIBRARY option to set MSVC ABI
* Fix some bugs in FindFLAC.cmake (closes#505)
The git version of libogg has a `--disable-crc` configure option which is
useful/necessary when fuzz testing. However, disabling libogg CRC is highly
undesirable when not fuzz testing so both need to be supported.
Also add configure script option checking.
Ogg: Introduce new functions ogg_sync_fseek(), ogg_ftell(), ogg_next_page(),
ogg_stream_next_page(), ogg_sync_last_page_before(), and
ogg_stream_unpack_page(). Use ogg_sync_* for ogg_read_first_page(). Bump libogg
version requirement from 1.1.3 to 1.3.0 as LibOgg 1.3.0 is required for
ogg_stream_pageout_fill() and ogg_stream_flush_fill().
Opus: Add opus support. Document added commands
SFC_(GET|SET)_ORIGINAL_SAMPLERATE. Added or extended tests ogg_opus_test,
compression_size_test, floating_point_test, lossy_comp_test, string_test,
external_libs_test. Change Opus to non-experimental.
i2ulaw_array() and i2alaw_array() fail to handle ptr [count] = INT_MIN
properly, leading to buffer underflow. INT_MIN is a special value
since - INT_MIN cannot be represented as int.
In this case round - INT_MIN to INT_MAX and proceed as usual.
f2ulaw_array() and f2alaw_array() fail to handle ptr [count] = NaN
properly, leading to null pointer dereference.
In this case, arbitrarily set the buffer value to 0.
This commit fixes#429 (CVE-2018-19661 and CVE-2018-19662) and
fixes#344 (CVE-2017-17456 and CVE-2017-17457).
While experimenting, determined that Vobis' analysis stops working on
non-normalized floating point input outside the range of [2000.0, -2000.0], and
seems to get worse the larger the magnitude. As the sdlcomp_test_float and
sldcomp_test_double tests work with non-normalized floats with a magnitude of
32000.0, this explains why they do not work for Vorbis.
This patch fixes the test by reducing the magnitude of the test data at encode
and restoring it at decode if it detects Vorbis, mantaining the same 32000.0
overall magnitude so the envelope values for the test are the same as they are
for other tests and format.
It is unknown at this time if the input range magnitude of 2000.0 is a
limitation of Vorbis or libvorbis.
* move SFC_SET_ADD_HEADER_PAD_CHUNK to deprecated section
* document SFC_GET/SET_CHANNEL_MAP_INFO
* docs: fix typo in command.html
* docs: fix typo in name of command
Closes: https://github.com/erikd/libsndfile/pull/422
