WMA Voice decoding context. More...

Data Fields
Global values specified in the stream header / extradata or used all over.
GetBitContext	gb
	packet bitreader.

int8_t	vbm_tree [25]
	converts VLC codes to frame type

int	spillover_bitsize
	number of bits used to specify spillover_nbits in the packet header = ceil(log2(ctx->block_align << 3))

int	history_nsamples
	number of samples in history for signal prediction (through ACB)

int	do_apf
	whether to apply the averaged projection filter (APF)

int	denoise_strength
	strength of denoising in Wiener filter [0-11]

int	denoise_tilt_corr
	Whether to apply tilt correction to the Wiener filter coefficients (postfilter)

int	dc_level
	Predicted amount of DC noise, based on which a DC removal filter is used.

int	lsps
	number of LSPs per frame [10 or 16]

int	lsp_q_mode
	defines quantizer defaults [0, 1]

int	lsp_def_mode
	defines different sets of LSP defaults [0, 1]

int	frame_lsp_bitsize
	size (in bits) of LSPs, when encoded per-frame (independent coding)

int	sframe_lsp_bitsize
	size (in bits) of LSPs, when encoded per superframe (residual coding)

int	min_pitch_val
	base value for pitch parsing code

int	max_pitch_val
	max value + 1 for pitch parsing

int	pitch_nbits
	number of bits used to specify the pitch value in the frame header

int	block_pitch_nbits
	number of bits used to specify the first block's pitch value

int	block_pitch_range
	range of the block pitch

int	block_delta_pitch_nbits
	number of bits used to specify the delta pitch between this and the last block's pitch value, used in all but first block

int	block_delta_pitch_hrange
	1/2 range of the delta (full range is from -this to +this-1)

uint16_t	block_conv_table [4]
	boundaries for block pitch unit/scale conversion

Packet values specified in the packet header or related to a packet.
A packet is considered to be a single unit of data provided to this decoder by the demuxer.
int	spillover_nbits
	number of bits of the previous packet's last superframe preceding this packet's first full superframe (useful for re-synchronization also)

int	has_residual_lsps
	if set, superframes contain one set of LSPs that cover all frames, encoded as independent and residual LSPs; if not set, each frame contains its own, fully independent, LSPs

int	skip_bits_next
	number of bits to skip at the next call to wmavoice_decode_packet() (since they're part of the previous superframe)

uint8_t	sframe_cache [SFRAME_CACHE_MAXSIZE+FF_INPUT_BUFFER_PADDING_SIZE]
	cache for superframe data split over multiple packets

int	sframe_cache_size
	set to >0 if we have data from an (incomplete) superframe from a previous packet that spilled over in the current packet; specifies the amount of bits in sframe_cache

PutBitContext	pb
	bitstream writer for sframe_cache

Frame and superframe values
Superframe and frame data - these can change from frame to frame, although some of them do in that case serve as a cache / history for the next frame or superframe.
double	prev_lsps [MAX_LSPS]
	LSPs of the last frame of the previous superframe.

int	last_pitch_val
	pitch value of the previous frame

int	last_acb_type
	frame type [0-2] of the previous frame

int	pitch_diff_sh16
	((cur_pitch_val - last_pitch_val) << 16) / MAX_FRAMESIZE

float	silence_gain
	set for use in blocks if ACB_TYPE_NONE

int	aw_idx_is_ext
	whether the AW index was encoded in 8 bits (instead of 6)

int	aw_pulse_range
	the range over which aw_pulse_set1() can apply the pulse, relative to the value in aw_first_pulse_off.

int	aw_n_pulses [2]
	number of AW-pulses in each block; note that this number can be negative (in which case it basically means "zero")

int	aw_first_pulse_off [2]
	index of first sample to which to apply AW-pulses, or -0xff if unset

int	aw_next_pulse_off_cache
	the position (relative to start of the second block) at which pulses should start to be positioned, serves as a cache for pitch-adaptive window pulses between blocks

int	frame_cntr
	current frame index [0 - 0xFFFE]; is only used for comfort noise in pRNG()

float	gain_pred_err [6]
	cache for gain prediction

float	excitation_history [MAX_SIGNAL_HISTORY]
	cache of the signal of previous superframes, used as a history for signal generation

float	synth_history [MAX_LSPS]
	see excitation_history

Postfilter values
Variables used for postfilter implementation, mostly history for smoothing and so on, and context variables for FFT/iFFT.
RDFTContext	rdft

RDFTContext	irdft
	contexts for FFT-calculation in the postfilter (for denoise filter)

DCTContext	dct

DCTContext	dst
	contexts for phase shift (in Hilbert transform, part of postfilter)

float	sin [511]

float	cos [511]
	8-bit cosine/sine windows over [-pi,pi] range

float	postfilter_agc
	gain control memory, used in adaptive_gain_control()

float	dcf_mem [2]
	DC filter history.

float	zero_exc_pf [MAX_SIGNAL_HISTORY+MAX_SFRAMESIZE]
	zero filter output (i.e.

float	denoise_filter_cache [MAX_FRAMESIZE]

int	denoise_filter_cache_size
	samples in denoise_filter_cache

float	tilted_lpcs_pf [0x80]
	aligned buffer for LPC tilting

float	denoise_coeffs_pf [0x80]
	aligned buffer for denoise coefficients

float	synth_filter_out_buf [0x80+MAX_LSPS_ALIGN16]
	aligned buffer for postfilter speech synthesis

Detailed Description

WMA Voice decoding context.

Definition at line 132 of file wmavoice.c.

Field Documentation

GetBitContext WMAVoiceContext::gb

packet bitreader.

During decoder init, it contains the extradata from the demuxer. During decoding, it contains packet data.

Definition at line 137 of file wmavoice.c.

Referenced by parse_packet_header(), synth_superframe(), wmavoice_decode_init(), and wmavoice_decode_packet().

int8_t WMAVoiceContext::vbm_tree[25]

converts VLC codes to frame type

Definition at line 141 of file wmavoice.c.

Referenced by check_bits_for_superframe(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::spillover_bitsize

number of bits used to specify spillover_nbits in the packet header = ceil(log2(ctx->block_align << 3))

Definition at line 143 of file wmavoice.c.

Referenced by parse_packet_header(), and wmavoice_decode_init().

int WMAVoiceContext::history_nsamples

number of samples in history for signal prediction (through ACB)

Definition at line 146 of file wmavoice.c.

Referenced by synth_frame(), synth_superframe(), wmavoice_decode_init(), and wmavoice_flush().

int WMAVoiceContext::do_apf

whether to apply the averaged projection filter (APF)

Definition at line 150 of file wmavoice.c.

Referenced by synth_frame(), synth_superframe(), wmavoice_decode_end(), wmavoice_decode_init(), and wmavoice_flush().

int WMAVoiceContext::denoise_strength

strength of denoising in Wiener filter [0-11]

Definition at line 152 of file wmavoice.c.

Referenced by calc_input_response(), and wmavoice_decode_init().

int WMAVoiceContext::denoise_tilt_corr

Whether to apply tilt correction to the Wiener filter coefficients (postfilter)

Definition at line 154 of file wmavoice.c.

Referenced by calc_input_response(), and wmavoice_decode_init().

int WMAVoiceContext::dc_level

Predicted amount of DC noise, based on which a DC removal filter is used.

Definition at line 156 of file wmavoice.c.

Referenced by postfilter(), and wmavoice_decode_init().

int WMAVoiceContext::lsps

number of LSPs per frame [10 or 16]

Definition at line 159 of file wmavoice.c.

Referenced by postfilter(), synth_block(), synth_frame(), synth_superframe(), wiener_denoise(), wmavoice_decode_init(), and wmavoice_flush().

int WMAVoiceContext::lsp_q_mode

defines quantizer defaults [0, 1]

Definition at line 160 of file wmavoice.c.

Referenced by synth_superframe(), and wmavoice_decode_init().

int WMAVoiceContext::lsp_def_mode

defines different sets of LSP defaults [0, 1]

Definition at line 161 of file wmavoice.c.

Referenced by synth_superframe(), and wmavoice_decode_init().

int WMAVoiceContext::frame_lsp_bitsize

size (in bits) of LSPs, when encoded per-frame (independent coding)

Definition at line 163 of file wmavoice.c.

Referenced by check_bits_for_superframe(), and wmavoice_decode_init().

int WMAVoiceContext::sframe_lsp_bitsize

size (in bits) of LSPs, when encoded per superframe (residual coding)

Definition at line 165 of file wmavoice.c.

Referenced by check_bits_for_superframe(), and wmavoice_decode_init().

int WMAVoiceContext::min_pitch_val

base value for pitch parsing code

Definition at line 168 of file wmavoice.c.

Referenced by kalman_smoothen(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::max_pitch_val

max value + 1 for pitch parsing

Definition at line 169 of file wmavoice.c.

Referenced by kalman_smoothen(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::pitch_nbits

number of bits used to specify the pitch value in the frame header

Definition at line 170 of file wmavoice.c.

Referenced by check_bits_for_superframe(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::block_pitch_nbits

number of bits used to specify the first block's pitch value

Definition at line 172 of file wmavoice.c.

Referenced by check_bits_for_superframe(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::block_pitch_range

range of the block pitch

Definition at line 174 of file wmavoice.c.

Referenced by synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::block_delta_pitch_nbits

number of bits used to specify the delta pitch between this and the last block's pitch value, used in all but first block

Definition at line 175 of file wmavoice.c.

Referenced by check_bits_for_superframe(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::block_delta_pitch_hrange

1/2 range of the delta (full range is from -this to +this-1)

Definition at line 179 of file wmavoice.c.

Referenced by synth_frame(), and wmavoice_decode_init().

uint16_t WMAVoiceContext::block_conv_table[4]

boundaries for block pitch unit/scale conversion

Definition at line 181 of file wmavoice.c.

Referenced by synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::spillover_nbits

number of bits of the previous packet's last superframe preceding this packet's first full superframe (useful for re-synchronization also)

Definition at line 193 of file wmavoice.c.

Referenced by parse_packet_header(), and wmavoice_decode_packet().

int WMAVoiceContext::has_residual_lsps

if set, superframes contain one set of LSPs that cover all frames, encoded as independent and residual LSPs; if not set, each frame contains its own, fully independent, LSPs

Definition at line 197 of file wmavoice.c.

Referenced by check_bits_for_superframe(), parse_packet_header(), and synth_superframe().

int WMAVoiceContext::skip_bits_next

number of bits to skip at the next call to wmavoice_decode_packet() (since they're part of the previous superframe)

Definition at line 202 of file wmavoice.c.

Referenced by wmavoice_decode_packet(), and wmavoice_flush().

uint8_t WMAVoiceContext::sframe_cache[SFRAME_CACHE_MAXSIZE+FF_INPUT_BUFFER_PADDING_SIZE]

cache for superframe data split over multiple packets

Definition at line 207 of file wmavoice.c.

Referenced by synth_superframe(), and wmavoice_decode_packet().

int WMAVoiceContext::sframe_cache_size

set to >0 if we have data from an (incomplete) superframe from a previous packet that spilled over in the current packet; specifies the amount of bits in sframe_cache

Definition at line 209 of file wmavoice.c.

Referenced by synth_superframe(), wmavoice_decode_packet(), and wmavoice_flush().

PutBitContext WMAVoiceContext::pb

bitstream writer for sframe_cache

Definition at line 214 of file wmavoice.c.

Referenced by wmavoice_decode_packet().

double WMAVoiceContext::prev_lsps[MAX_LSPS]

LSPs of the last frame of the previous superframe.

Definition at line 225 of file wmavoice.c.

Referenced by synth_superframe(), wmavoice_decode_init(), and wmavoice_flush().

int WMAVoiceContext::last_pitch_val

pitch value of the previous frame

Definition at line 227 of file wmavoice.c.

Referenced by synth_block_fcb_acb(), synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::last_acb_type

frame type [0-2] of the previous frame

Definition at line 228 of file wmavoice.c.

Referenced by synth_frame(), and wmavoice_decode_init().

int WMAVoiceContext::pitch_diff_sh16

((cur_pitch_val - last_pitch_val) << 16) / MAX_FRAMESIZE

Definition at line 229 of file wmavoice.c.

Referenced by synth_block_fcb_acb(), and synth_frame().

float WMAVoiceContext::silence_gain

set for use in blocks if ACB_TYPE_NONE

Definition at line 231 of file wmavoice.c.

Referenced by synth_block_hardcoded(), and synth_frame().

int WMAVoiceContext::aw_idx_is_ext

whether the AW index was encoded in 8 bits (instead of 6)

Definition at line 233 of file wmavoice.c.

Referenced by aw_parse_coords(), and aw_pulse_set1().

int WMAVoiceContext::aw_pulse_range

the range over which aw_pulse_set1() can apply the pulse, relative to the value in aw_first_pulse_off.

The exact position of the first AW-pulse is within [pulse_off, pulse_off + this], and depends on bitstream values; [16 or 24]

Definition at line 235 of file wmavoice.c.

Referenced by aw_parse_coords(), aw_pulse_set1(), and aw_pulse_set2().

int WMAVoiceContext::aw_n_pulses[2]

number of AW-pulses in each block; note that this number can be negative (in which case it basically means "zero")

Definition at line 241 of file wmavoice.c.

Referenced by aw_parse_coords(), aw_pulse_set1(), and aw_pulse_set2().

int WMAVoiceContext::aw_first_pulse_off[2]

index of first sample to which to apply AW-pulses, or -0xff if unset

Definition at line 244 of file wmavoice.c.

Referenced by aw_parse_coords(), aw_pulse_set1(), and aw_pulse_set2().

int WMAVoiceContext::aw_next_pulse_off_cache

the position (relative to start of the second block) at which pulses should start to be positioned, serves as a cache for pitch-adaptive window pulses between blocks

Definition at line 246 of file wmavoice.c.

Referenced by aw_pulse_set2().

int WMAVoiceContext::frame_cntr

current frame index [0 - 0xFFFE]; is only used for comfort noise in pRNG()

Definition at line 252 of file wmavoice.c.

Referenced by synth_block_hardcoded(), and synth_frame().

float WMAVoiceContext::gain_pred_err[6]

cache for gain prediction

Definition at line 254 of file wmavoice.c.

Referenced by synth_block_fcb_acb(), synth_block_hardcoded(), and wmavoice_flush().

float WMAVoiceContext::excitation_history[MAX_SIGNAL_HISTORY]

cache of the signal of previous superframes, used as a history for signal generation