NRF_EDGEAI_DECLARE_FEATURE_FUNCTION

#define NRF_EDGEAI_DECLARE_FEATURE_FUNCTION	(		feature_name,
			input_type,
			feature_type )

Value:

    size32_t nrf_edgeai_feature_##feature_name(input_type*                     p_input,        \
                                               size32_t                        num,            \
                                               feature_type*                   p_features,     \
                                               nrf_edgeai_features_mask_t      feature_mask,   \
                                               void*                           p_pipeline_ctx, \
                                               nrf_edgeai_feature_get_arg_cb_t get_argument,   \
                                               void*                           p_argument_ctx)

Macro to declare time-domain DSP feature extraction functions for various data types.

All declared functions have the following prototype:

size32_t nrf_edgeai_feature_<feature_name>(
    <input_type>*                    p_input,
    size32_t                      num,
    <feature_type>*                  p_features,
    nrf_edgeai_features_mask_t        feature_mask,
    void*                            p_pipeline_ctx,
    nrf_edgeai_feature_get_arg_cb_t   get_argument,
    void*                            p_argument_ctx
);

where:

• <feature_name> : Name of the feature (e.g., mean_i8, std_f32, etc.)
• <input_type> : Input data type (int8_t, int16_t, flt32_t)
• <feature_type> : Output feature data type (int16_t, int32_t, flt32_t)

These functions extract one or more time-domain features from the input signal buffer, storing the results in the output feature buffer. The feature_mask parameter allows selective extraction of features. The pipeline context and argument callback provide additional configuration or arguments for feature extraction.

Parameters of the expanded function:

• p_input Pointer to the input signal buffer.
• num Number of input samples.
• p_features Pointer to the output feature buffer.
• feature_mask Mask specifying which features to extract.
• p_pipeline_ctx Pointer to the pipeline context (may be NULL).
• get_argument Callback to get feature-specific arguments (may be NULL).
• p_argument_ctx Pointer to argument context for the callback (may be NULL).
• Return: Number of features extracted and written to p_features.

Note

• Use NRF_EDGEAI_DECLARE_FEATURE_FUNCTION_I8, NRF_EDGEAI_DECLARE_FEATURE_FUNCTION_I16, or NRF_EDGEAI_DECLARE_FEATURE_FUNCTION_F32 to declare feature functions for int8, int16, or float32 input types, respectively.