Software "pseudo-float": a real number represented as a signed 16-bit mantissa plus a 16-bit exponent (value = mantissa * 2^exponent), computed with integer multiply/shift/compare only – no hardware floating-point instructions. This generalizes speexdsp's classic fixed-point trick of tracking a handful of wide-dynamic-range control variables as a normalized mantissa+exponent pair into a full, operator-overloaded numeric type usable as a drop-in replacement for float (see MDFFixedPoint in MDFEchoCancellation.h, which uses this class as the element type for the MDF echo canceller's sample/spectrum arrays and internal state, so the whole algorithm runs without native float arithmetic). More...

#include <soft_float_t.h>

Public Member Functions
constexpr	soft_float_t ()=default
	soft_float_t (double d)
	soft_float_t (float f)
	soft_float_t (int i)
constexpr	soft_float_t (int16_t mantissa, int16_t exponent)
int16_t	exponent () const
int16_t	mantissa () const
	operator float () const
	operator int () const
	truncates toward zero, like (int)someFloat
bool	operator!= (double o) const
bool	operator!= (float o) const
bool	operator!= (int o) const
bool	operator!= (soft_float_t o) const
soft_float_t	operator* (double o) const
soft_float_t	operator* (float o) const
soft_float_t	operator* (int o) const
soft_float_t	operator* (soft_float_t o) const
soft_float_t &	operator*= (soft_float_t o)
soft_float_t	operator+ (double o) const
soft_float_t	operator+ (float o) const
soft_float_t	operator+ (int o) const
soft_float_t	operator+ (soft_float_t o) const
soft_float_t &	operator+= (soft_float_t o)
soft_float_t	operator- () const
soft_float_t	operator- (double o) const
soft_float_t	operator- (float o) const
soft_float_t	operator- (int o) const
soft_float_t	operator- (soft_float_t o) const
soft_float_t &	operator-= (soft_float_t o)
soft_float_t	operator/ (double o) const
soft_float_t	operator/ (float o) const
soft_float_t	operator/ (int o) const
soft_float_t	operator/ (soft_float_t o) const
soft_float_t &	operator/= (soft_float_t o)
bool	operator< (double o) const
bool	operator< (float o) const
bool	operator< (int o) const
bool	operator< (soft_float_t o) const
bool	operator<= (double o) const
bool	operator<= (float o) const
bool	operator<= (int o) const
bool	operator<= (soft_float_t o) const
bool	operator== (double o) const
bool	operator== (float o) const
bool	operator== (int o) const
bool	operator== (soft_float_t o) const
bool	operator> (double o) const
bool	operator> (float o) const
bool	operator> (int o) const
bool	operator> (soft_float_t o) const
bool	operator>= (double o) const
bool	operator>= (float o) const
bool	operator>= (int o) const
bool	operator>= (soft_float_t o) const
int16_t	scale (int16_t sample) const
	scales a 16-bit PCM sample by this factor
int24_t	scale (int24_t sample) const
	scales a 24-bit PCM sample by this factor
int32_t	scale (int32_t sample) const
	scales a 32-bit PCM sample by this factor
int16_t	toInt16 () const
	converts to a 16-bit PCM sample (-32768..32767)
int24_t	toInt24 () const
	converts to a 24-bit PCM sample (-8388608..8388607)
int32_t	toInt32 () const
	converts to a 32-bit PCM sample (-2147483648..2147483647)

Static Public Member Functions
static soft_float_t	fromInt16 (int16_t sample)
	builds a soft_float_t from a 16-bit PCM sample (-32768..32767)
static soft_float_t	fromInt24 (int24_t sample)
	builds a soft_float_t from a 24-bit PCM sample (-8388608..8388607)
static soft_float_t	fromInt32 (int32_t sample)
	builds a soft_float_t from a 32-bit PCM sample (-2147483648..2147483647)

Detailed Description

Software "pseudo-float": a real number represented as a signed 16-bit mantissa plus a 16-bit exponent (value = mantissa * 2^exponent), computed with integer multiply/shift/compare only – no hardware floating-point instructions. This generalizes speexdsp's classic fixed-point trick of tracking a handful of wide-dynamic-range control variables as a normalized mantissa+exponent pair into a full, operator-overloaded numeric type usable as a drop-in replacement for float (see MDFFixedPoint in MDFEchoCancellation.h, which uses this class as the element type for the MDF echo canceller's sample/spectrum arrays and internal state, so the whole algorithm runs without native float arithmetic).

Unlike a plain fixed Q-format (a fixed number of fractional bits, which only covers a fixed range/precision trade-off and would over/underflow badly across this algorithm's actual value range – audio samples, FFT magnitudes in the thousands, and adaptation-rate fractions near zero, all flowing through the same arrays), this type renormalizes on every operation, so it tolerates the same wide dynamic range as float, at roughly the precision of a 16-bit mantissa (a relative error on the order of 2^-15, i.e. about 4-5 significant decimal digits – comparable to a compressed/half-precision float).

Note: Comparison operators (<, >, etc.) convert both sides to float and compare natively – they're scalar, infrequent control-flow checks, not the per-element array math this class exists to keep off the FPU, so trading a little purity there for simpler, more obviously correct code is a reasonable trade. Arithmetic that mixes a soft_float_t with a plain float/double literal (e.g. 0.7f * pseudoFloatValue) also still executes that specific operation using native float arithmetic via the implicit conversions below – avoiding all native float instructions everywhere would require rewriting every numeric literal in the algorithm as a soft_float_t, which isn't done here. What this class does guarantee is that soft_float_t-to-soft_float_t arithmetic – the tight inner loops over whole arrays (FFT bins, filter weights, correlations) that dominate the algorithm's cost – runs on integer mantissa/exponent math throughout.

Constructor & Destructor Documentation

◆ soft_float_t() [1/5]

soft_float_t ( )

constexprdefault

◆ soft_float_t() [2/5]

soft_float_t	(	int16_t	mantissa,
		int16_t	exponent )

inlineconstexpr

◆ soft_float_t() [3/5]

soft_float_t ( float f )

inline

◆ soft_float_t() [4/5]

soft_float_t ( double d )

inline

◆ soft_float_t() [5/5]

soft_float_t ( int i )

inline

Member Function Documentation

◆ exponent()

int16_t exponent ( ) const

inline

◆ fromInt16()

soft_float_t fromInt16 ( int16_t sample )

inlinestatic

builds a soft_float_t from a 16-bit PCM sample (-32768..32767)

◆ fromInt24()

soft_float_t fromInt24 ( int24_t sample )

inlinestatic

builds a soft_float_t from a 24-bit PCM sample (-8388608..8388607)

◆ fromInt32()

soft_float_t fromInt32 ( int32_t sample )

inlinestatic

builds a soft_float_t from a 32-bit PCM sample (-2147483648..2147483647)

◆ mantissa()

int16_t mantissa ( ) const

inline

◆ operator float()

operator float ( ) const

inline

◆ operator int()

operator int ( ) const

inlineexplicit

truncates toward zero, like (int)someFloat

◆ operator!=() [1/4]

bool operator!= ( double o ) const

inline

◆ operator!=() [2/4]

bool operator!= ( float o ) const

inline

◆ operator!=() [3/4]

bool operator!= ( int o ) const

inline

◆ operator!=() [4/4]

bool operator!= ( soft_float_t o ) const

inline

◆ operator*() [1/4]

soft_float_t operator* ( double o ) const

inline

◆ operator*() [2/4]

soft_float_t operator* ( float o ) const

inline

◆ operator*() [3/4]

soft_float_t operator* ( int o ) const

inline

◆ operator*() [4/4]

soft_float_t operator* ( soft_float_t o ) const

inline

◆ operator*=()

soft_float_t & operator*= ( soft_float_t o )

inline

◆ operator+() [1/4]

soft_float_t operator+ ( double o ) const

inline

◆ operator+() [2/4]

soft_float_t operator+ ( float o ) const

inline

◆ operator+() [3/4]

soft_float_t operator+ ( int o ) const

inline

◆ operator+() [4/4]

soft_float_t operator+ ( soft_float_t o ) const

inline

◆ operator+=()

soft_float_t & operator+= ( soft_float_t o )

inline

◆ operator-() [1/5]

soft_float_t operator- ( ) const

inline

◆ operator-() [2/5]

soft_float_t operator- ( double o ) const

inline

◆ operator-() [3/5]

soft_float_t operator- ( float o ) const

inline

◆ operator-() [4/5]

soft_float_t operator- ( int o ) const

inline

◆ operator-() [5/5]

soft_float_t operator- ( soft_float_t o ) const

inline

◆ operator-=()

soft_float_t & operator-= ( soft_float_t o )

inline

◆ operator/() [1/4]

soft_float_t operator/ ( double o ) const

inline

◆ operator/() [2/4]

soft_float_t operator/ ( float o ) const

inline

◆ operator/() [3/4]

soft_float_t operator/ ( int o ) const

inline

◆ operator/() [4/4]

soft_float_t operator/ ( soft_float_t o ) const

inline

◆ operator/=()

soft_float_t & operator/= ( soft_float_t o )

inline

◆ operator<() [1/4]

bool operator< ( double o ) const

inline

◆ operator<() [2/4]

bool operator< ( float o ) const

inline

◆ operator<() [3/4]

bool operator< ( int o ) const

inline

◆ operator<() [4/4]

bool operator< ( soft_float_t o ) const

inline

◆ operator<=() [1/4]

bool operator<= ( double o ) const

inline

◆ operator<=() [2/4]

bool operator<= ( float o ) const

inline

◆ operator<=() [3/4]

bool operator<= ( int o ) const

inline

◆ operator<=() [4/4]

bool operator<= ( soft_float_t o ) const

inline

◆ operator==() [1/4]

bool operator== ( double o ) const

inline

◆ operator==() [2/4]

bool operator== ( float o ) const

inline

◆ operator==() [3/4]

bool operator== ( int o ) const

inline

◆ operator==() [4/4]

bool operator== ( soft_float_t o ) const

inline

◆ operator>() [1/4]

bool operator> ( double o ) const

inline

◆ operator>() [2/4]

bool operator> ( float o ) const

inline

◆ operator>() [3/4]

bool operator> ( int o ) const

inline

◆ operator>() [4/4]

bool operator> ( soft_float_t o ) const

inline

◆ operator>=() [1/4]

bool operator>= ( double o ) const

inline

◆ operator>=() [2/4]

bool operator>= ( float o ) const

inline

◆ operator>=() [3/4]

bool operator>= ( int o ) const

inline

◆ operator>=() [4/4]

bool operator>= ( soft_float_t o ) const

inline

◆ scale() [1/3]

int16_t scale ( int16_t sample ) const

inline

scales a 16-bit PCM sample by this factor

◆ scale() [2/3]

int24_t scale ( int24_t sample ) const

inline

scales a 24-bit PCM sample by this factor

◆ scale() [3/3]

int32_t scale ( int32_t sample ) const

inline

scales a 32-bit PCM sample by this factor

◆ toInt16()

int16_t toInt16 ( ) const

inline

converts to a 16-bit PCM sample (-32768..32767)

◆ toInt24()

int24_t toInt24 ( ) const

inline

converts to a 24-bit PCM sample (-8388608..8388607)

◆ toInt32()

int32_t toInt32 ( ) const

inline

converts to a 32-bit PCM sample (-2147483648..2147483647)

The documentation for this class was generated from the following file:

src/AudioTools/CoreAudio/AudioBasic/soft_float_t.h

Public Member Functions

Static Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ soft_float_t() [1/5]

◆ soft_float_t() [2/5]

◆ soft_float_t() [3/5]

◆ soft_float_t() [4/5]

◆ soft_float_t() [5/5]

Member Function Documentation

◆ exponent()

◆ fromInt16()

◆ fromInt24()

◆ fromInt32()

◆ mantissa()

◆ operator float()

◆ operator int()

◆ operator!=() [1/4]

◆ operator!=() [2/4]

◆ operator!=() [3/4]

◆ operator!=() [4/4]

◆ operator*() [1/4]

◆ operator*() [2/4]

◆ operator*() [3/4]

◆ operator*() [4/4]

◆ operator*=()

◆ operator+() [1/4]

◆ operator+() [2/4]

◆ operator+() [3/4]

◆ operator+() [4/4]

◆ operator+=()

◆ operator-() [1/5]

◆ operator-() [2/5]

◆ operator-() [3/5]

◆ operator-() [4/5]

◆ operator-() [5/5]

◆ operator-=()

◆ operator/() [1/4]

◆ operator/() [2/4]

◆ operator/() [3/4]

◆ operator/() [4/4]

◆ operator/=()

◆ operator<() [1/4]

◆ operator<() [2/4]

◆ operator<() [3/4]

◆ operator<() [4/4]

◆ operator<=() [1/4]

◆ operator<=() [2/4]

◆ operator<=() [3/4]

◆ operator<=() [4/4]

◆ operator==() [1/4]

◆ operator==() [2/4]

◆ operator==() [3/4]

◆ operator==() [4/4]

◆ operator>() [1/4]

◆ operator>() [2/4]

◆ operator>() [3/4]

◆ operator>() [4/4]

◆ operator>=() [1/4]

◆ operator>=() [2/4]

◆ operator>=() [3/4]

◆ operator>=() [4/4]

◆ scale() [1/3]

◆ scale() [2/3]

◆ scale() [3/3]

◆ toInt16()

◆ toInt24()

◆ toInt32()