Go to the source code of this file.
|
| int32_t | sofm_exp_approx (int32_t x) |
| | Calculates exponent function exp(x) = e^x with accurate and efficient technique that includes range reduction operations, approximation with Taylor series, and reconstruction operations to compensate the range reductions.
|
| int32_t | sofm_exp_fixed (int32_t x) |
| | Calculated exponent function exp(x) = e^x by using sofm_exp_approx().
|
| int32_t | sofm_db2lin_fixed (int32_t db) |
| | Converts a decibels value to liner amplitude lin = 10^(db/20) value with optimized equation exp(db * log(10)/20).
|
◆ EXPONENTIAL_GENERIC
| #define EXPONENTIAL_GENERIC 1 |
◆ SOFM_DB2LIN_INPUT_MAX
| #define SOFM_DB2LIN_INPUT_MAX 1111097957 |
◆ SOFM_EXP_FIXED_INPUT_MAX
| #define SOFM_EXP_FIXED_INPUT_MAX 1023359037 |
◆ sofm_db2lin_fixed()
| int32_t sofm_db2lin_fixed |
( |
int32_t | db | ) |
|
Converts a decibels value to liner amplitude lin = 10^(db/20) value with optimized equation exp(db * log(10)/20).
Note that due to range limitation of sofm_exp_fixed() the output saturates to maximum with about +66 dB input.
- Parameters
-
| db | Decibels value in Q8.24 format, from -128 to +66.226 |
- Returns
- Linear value in Q12.20 format, from 3.9811e-07 to 2048
◆ sofm_exp_approx()
| int32_t sofm_exp_approx |
( |
int32_t | x | ) |
|
Calculates exponent function exp(x) = e^x with accurate and efficient technique that includes range reduction operations, approximation with Taylor series, and reconstruction operations to compensate the range reductions.
- Parameters
-
| x | The input argument as Q4.28 from -8 to +8 |
- Returns
- The calculated e^x value as Q13.19 from 3.3546e-04 to 2981.0
◆ sofm_exp_fixed()
| int32_t sofm_exp_fixed |
( |
int32_t | x | ) |
|
Calculated exponent function exp(x) = e^x by using sofm_exp_approx().
The input range for large arguments is internally reduced to -8 to +8 with rule exp(x) = exp(x/2) * exp(x/2) and reconstructed back. This function is essentially a wrapper for compatibility with existing usage of exp() function and Q-format choice. Note that the return value saturates to INT32_MAX with input arguments larger than 7.6246.
- Parameters
-
| x | The input argument as Q5.27 from -16 to +16 |
- Returns
- The calculated e^x value as Q12.20
