FIRconv

0.0.3 verified Mon Apr 27 auth: no python

Python library for real-time convolution, primarily for auralization. Current version 0.0.3, early development with infrequent releases.

pip install firconv

Common errors

error ModuleNotFoundError: No module named 'firconv' ↓

cause Library is not installed or you are using the wrong package name (e.g., 'pyfir', 'fir').

fix

Run: pip install firconv

error AttributeError: module 'firconv' has no attribute 'FIRconv' ↓

cause You used 'import firconv' and tried firconv.FIRconv, but FIRconv is a class directly importable.

fix

Use: from firconv import FIRconv

error ValueError: Input must be float32 ↓

cause Passing int16 or float64 arrays to FIRconv.process().

fix

Convert input: input_signal = input_signal.astype(np.float32)

Warnings

breaking The API may change drastically before 1.0. No backward compatibility guarantees. ↓

fix Pin version and review release notes before upgrading.

gotcha Audio data must be float32 numpy arrays. Other types (int16, float64) cause silent errors or crashes. ↓

fix Ensure input arrays are dtype=np.float32.

gotcha Impulse response length must be a power of two or at least aligned with block size. Non-power-of-two IRs may produce artifacts. ↓

fix Pad IR to a power of two if necessary.

Imports

FIRconv
wrong
```
import FIRconv
```
correct
```
from firconv import FIRconv
```
FIRconv is a class, not a module; direct import of the class is correct.

PartitionedConvolution

wrong

from firconv.core import PartitionedConvolution

correct

from firconv import PartitionedConvolution

PartitionedConvolution is exposed at package level.

Quickstart

Basic usage: create FIRconv instance with an impulse response and process audio blocks.

from firconv import FIRconv
import numpy as np

# Create a simple impulse response (e.g., 1024 samples)
ir = np.random.randn(1024).astype(np.float32)

# Initialize FIRconv with the IR
conv = FIRconv(ir)

# Process a block of audio
input_signal = np.random.randn(4096).astype(np.float32)
output = conv.process(input_signal)

# For real-time streaming, use process() repeatedly