File uq15_to_f32_core.h¶

FileList > inc > uq15_to_f32 > uq15_to_f32_core.h

UQ15 (offset-binary uint16) to float converter. More...

#include "clib_common.h"
#include "jm_perf.h"

Classes¶

Type	Name
struct	uq15_to_f32_state_t UQ15ToF32 state.

Public Functions¶

Type	Name
uq15_to_f32_state_t *	uq15_to_f32_create (float scale) Create a uq15_to_f32 instance.
void	uq15_to_f32_destroy (uq15_to_f32_state_t * state) Destroy a uq15_to_f32 instance and release all memory.
void	uq15_to_f32_reset (uq15_to_f32_state_t * state) Reset uq15_to_f32 to its post-create state.
JM_FORCEINLINE JM_HOT float	uq15_to_f32_step (const uq15_to_f32_state_t * state, uint16_t x) Process one input sample.
void	uq15_to_f32_steps (uq15_to_f32_state_t * state, const uint16_t * input, float * output, size_t n) Process a block of UQ15 samples to float32.

Detailed Description¶

Decodes an offset-binary uint16 (UQ15) sample back to a normalised float by removing the +32768 bias and dividing by scale:

x̂ = ((int32_t)u - 32768) * (1 / scale)

This is the exact inverse of F32ToUQ15 with the same scale. The bias removal uses int32_t arithmetic to avoid signed overflow for the u=0 (full-negative) case. The inverse scale is pre-computed at construction time so the step path is a single subtract and multiply.

Lifecycle: create -> (step / steps / reset)* -> destroy

>>> from doppler.cvt import UQ15ToF32
>>> import numpy as np
>>> obj = UQ15ToF32(scale=32768.0)
>>> float(obj.step(32768))
0.0
>>> float(obj.step(0))
-1.0
>>> x = np.array([0, 32768, 65535], dtype=np.uint16)
>>> [round(v, 6) for v in obj.steps(x).tolist()]
[-1.0, 0.0, 0.999969]

Public Functions Documentation¶

function uq15_to_f32_create¶

Create a uq15_to_f32 instance.

uq15_to_f32_state_t * uq15_to_f32_create (
    float scale
)

Pre-computes iscale = 1.0f / scale so the hot step path is a single subtract and multiply.

Parameters:

scale Denominator applied after offset-binary bias removal (default: 32768.0f). Use 32768.0 to recover normalised [-1, +1] floats from UQ15 data written by F32ToUQ15. Must be > 0; returns NULL otherwise.

Returns:

Heap-allocated state, or NULL on invalid args or allocation failure.

Note:

Caller must call uq15_to_f32_destroy() when done.

function uq15_to_f32_destroy¶

Destroy a uq15_to_f32 instance and release all memory.

void uq15_to_f32_destroy (
    uq15_to_f32_state_t * state
)

Parameters:

state May be NULL.

function uq15_to_f32_reset¶

Reset uq15_to_f32 to its post-create state.

void uq15_to_f32_reset (
    uq15_to_f32_state_t * state
)

No mutable state exists beyond the immutable iscale; reset is a no-op provided for lifecycle symmetry with other converters.

Parameters:

state Must be non-NULL.

function uq15_to_f32_step¶

Process one input sample.

JM_FORCEINLINE  JM_HOT float uq15_to_f32_step (
    const uq15_to_f32_state_t * state,
    uint16_t x
)

Computes `((int32\_t)x - 32768) \* iscale. The int32\_t cast prevents signed overflow whenx` is 0 (which yields -32768 after bias removal).

Parameters:

state Must be non-NULL.
x UQ15 offset-binary uint16 sample: 0x0000 → -1.0f, 0x8000 → 0.0f, 0xFFFF → +32767/32768.

Returns:

Decoded float sample in [-1.0, ~+1.0).

function uq15_to_f32_steps¶

Process a block of UQ15 samples to float32.

void uq15_to_f32_steps (
    uq15_to_f32_state_t * state,
    const uint16_t * input,
    float * output,
    size_t n
)

Applies step() to every element. State is not mutated (no clipped flag). Accepts an optional pre-allocated output array; allocates a fresh one when output is NULL.

Parameters:

state Must be non-NULL.
input Input uint16 offset-binary array; must contain at least n elements.
output Output float32 array; must contain at least n elements.
n Number of samples to process.

>>> from doppler.cvt import UQ15ToF32
>>> import numpy as np
>>> UQ15ToF32().steps(np.array([0, 32768], dtype=np.uint16)).tolist()
[-1.0, 0.0]

The documentation for this class was generated from the following file native/inc/uq15_to_f32/uq15_to_f32_core.h