ladybird/Userland/Libraries/LibAudio/FlacLoader.h
kleines Filmröllchen 111fd1e5fe LibAudio: Optimize FlacLoader by avoiding allocations
- Pre-allocate and reuse sample decompression buffers. In many FLAC
  files, the amount of samples per frame is either constant or the
  largest frame will be hit within the first couple of frames. Also,
  during audio output, we need to move and combine the samples from the
  decompression buffers into the final output buffers anyways. Avoiding
  the reallocation of these large buffers provides an improvement from
  16x to 18x decode speed on strongly compressed but otherwise usual
  input.
- Leave a FIXME for a similar improvement that can be made in the
  residual decoder.
- Pre-allocate audio chunks if frame size is known.
- Use reasonable inline capacities in several places where we know the
  maximum or usual capacity needed.
2023-07-06 19:13:36 +01:00

120 lines
6.3 KiB
C++

/*
* Copyright (c) 2021, kleines Filmröllchen <filmroellchen@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#pragma once
#include "FlacTypes.h"
#include "Loader.h"
#include <AK/BitStream.h>
#include <AK/Error.h>
#include <AK/Span.h>
#include <AK/Types.h>
namespace Audio {
ALWAYS_INLINE u8 frame_channel_type_to_channel_count(FlacFrameChannelType channel_type);
// Sign-extend an arbitrary-size signed number to 64 bit signed
ALWAYS_INLINE i64 sign_extend(u32 n, u8 size);
// Decodes the sign representation method used in Rice coding.
// Numbers alternate between positive and negative: 0, 1, -1, 2, -2, 3, -3, 4, -4, 5, -5, ...
ALWAYS_INLINE i32 rice_to_signed(u32 x);
// decoders
// read a UTF-8 encoded number, even if it is not a valid codepoint
ALWAYS_INLINE ErrorOr<u64> read_utf8_char(BigEndianInputBitStream& input);
// decode a single number encoded with exponential golomb encoding of the specified order
ALWAYS_INLINE ErrorOr<i32> decode_unsigned_exp_golomb(u8 order, BigEndianInputBitStream& bit_input);
// Loader for the Free Lossless Audio Codec (FLAC)
// This loader supports all audio features of FLAC, although audio from more than two channels is discarded.
// The loader currently supports the STREAMINFO, PADDING, and SEEKTABLE metadata blocks.
// See: https://xiph.org/flac/documentation_format_overview.html
// https://xiph.org/flac/format.html (identical to IETF draft version 2)
// https://datatracker.ietf.org/doc/html/draft-ietf-cellar-flac-02 (all section numbers refer to this specification)
// https://datatracker.ietf.org/doc/html/draft-ietf-cellar-flac-03 (newer IETF draft that uses incompatible numberings and names)
class FlacLoaderPlugin : public LoaderPlugin {
public:
explicit FlacLoaderPlugin(NonnullOwnPtr<SeekableStream> stream);
virtual ~FlacLoaderPlugin() override = default;
static bool sniff(SeekableStream& stream);
static ErrorOr<NonnullOwnPtr<LoaderPlugin>, LoaderError> create(NonnullOwnPtr<SeekableStream>);
virtual ErrorOr<Vector<FixedArray<Sample>>, LoaderError> load_chunks(size_t samples_to_read_from_input) override;
virtual MaybeLoaderError reset() override;
virtual MaybeLoaderError seek(int sample_index) override;
virtual int loaded_samples() override { return static_cast<int>(m_loaded_samples); }
virtual int total_samples() override { return static_cast<int>(m_total_samples); }
virtual u32 sample_rate() override { return m_sample_rate; }
virtual u16 num_channels() override { return m_num_channels; }
virtual DeprecatedString format_name() override { return "FLAC (.flac)"; }
virtual PcmSampleFormat pcm_format() override { return m_sample_format; }
bool is_fixed_blocksize_stream() const { return m_min_block_size == m_max_block_size; }
bool sample_count_unknown() const { return m_total_samples == 0; }
private:
MaybeLoaderError initialize();
MaybeLoaderError parse_header();
// Either returns the metadata block or sets error message.
// Additionally, increments m_data_start_location past the read meta block.
ErrorOr<FlacRawMetadataBlock, LoaderError> next_meta_block(BigEndianInputBitStream& bit_input);
// Fetches and returns the next FLAC frame.
LoaderSamples next_frame();
// Helper of next_frame that fetches a sub frame's header
ErrorOr<FlacSubframeHeader, LoaderError> next_subframe_header(BigEndianInputBitStream& bit_input, u8 channel_index);
// Helper of next_frame that decompresses a subframe
ErrorOr<void, LoaderError> parse_subframe(Vector<i64>& samples, FlacSubframeHeader& subframe_header, BigEndianInputBitStream& bit_input);
// Subframe-internal data decoders (heavy lifting)
ErrorOr<Vector<i64>, LoaderError> decode_fixed_lpc(FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
ErrorOr<Vector<i64>, LoaderError> decode_verbatim(FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
ErrorOr<void, LoaderError> decode_custom_lpc(Vector<i64>& decoded, FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
MaybeLoaderError decode_residual(Vector<i64>& decoded, FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
// decode a single rice partition that has its own rice parameter
ALWAYS_INLINE ErrorOr<Vector<i64>, LoaderError> decode_rice_partition(u8 partition_type, u32 partitions, u32 partition_index, FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
MaybeLoaderError load_seektable(FlacRawMetadataBlock&);
// Note that failing to read a Vorbis comment block is not treated as an error of the FLAC loader, since metadata is optional.
void load_vorbis_comment(FlacRawMetadataBlock&);
MaybeLoaderError load_picture(FlacRawMetadataBlock&);
// Converters for special coding used in frame headers
ALWAYS_INLINE ErrorOr<u32, LoaderError> convert_sample_count_code(u8 sample_count_code);
ALWAYS_INLINE ErrorOr<u32, LoaderError> convert_sample_rate_code(u8 sample_rate_code);
ALWAYS_INLINE ErrorOr<u8, LoaderError> convert_bit_depth_code(u8 bit_depth_code);
bool should_insert_seekpoint_at(u64 sample_index) const;
// Data obtained directly from the FLAC metadata: many values have specific bit counts
u32 m_sample_rate { 0 }; // 20 bit
u8 m_num_channels { 0 }; // 3 bit
u8 m_bits_per_sample { 0 }; // 5 bits for the integer bit depth
// Externally visible format; the smallest integer format that's larger than the precise bit depth.
PcmSampleFormat m_sample_format;
// Blocks are units of decoded audio data
u16 m_min_block_size { 0 };
u16 m_max_block_size { 0 };
// Frames are units of encoded audio data, both of these are 24-bit
u32 m_min_frame_size { 0 }; // 24 bit
u32 m_max_frame_size { 0 }; // 24 bit
u64 m_total_samples { 0 }; // 36 bit
u8 m_md5_checksum[128 / 8]; // 128 bit (!)
size_t m_loaded_samples { 0 };
// keep track of the start of the data in the FLAC stream to seek back more easily
u64 m_data_start_location { 0 };
Optional<FlacFrameHeader> m_current_frame;
u64 m_current_sample_or_frame { 0 };
SeekTable m_seektable;
// Keep around a few temporary buffers whose allocated space can be reused.
// This is an empirical optimization since allocations and deallocations take a lot of time in the decoder.
mutable Vector<Vector<i64>, 2> m_subframe_buffers;
};
}