ComfyUI Node: MW Audio Recorder

MW Audio Recorder:

The AudioRecorderDR node is designed to facilitate audio recording and processing within the ComfyUI environment, specifically tailored for AI artists who may not have extensive technical expertise. This node captures audio input, processes it to reduce noise, and outputs a clean audio waveform. It employs advanced techniques such as spectral gating and dynamic gain normalization to enhance audio quality. The node is particularly beneficial for those looking to integrate high-quality audio into their AI-driven projects, offering a seamless way to manage audio recording and cleaning without requiring deep technical knowledge.

MW Audio Recorder Input Parameters:

record_sec

This parameter specifies the duration of the audio recording in seconds. It determines how long the recording session will last, directly impacting the length of the captured audio. The minimum value is 1 second, and the default is set to 5 seconds, allowing for flexibility in capturing short to moderately long audio clips.

sample_rate

The sample_rate parameter defines the number of samples per second in the audio recording, affecting the audio quality and file size. Available options are "16000", "44100", and "48000", with "48000" as the default. Higher sample rates provide better audio fidelity but result in larger file sizes.

n_fft

This parameter sets the number of Fast Fourier Transform (FFT) points used in the spectral analysis, influencing the frequency resolution of the audio processing. It ranges from 512 to 4096, with a default of 2048. A higher n_fft value offers finer frequency detail but requires more computational resources.

sensitivity

The sensitivity parameter adjusts the threshold for noise reduction, affecting how aggressively noise is filtered from the audio. It ranges from 0.1 to 3.0, with a default of 1.2. Higher sensitivity values result in more noise being removed, which can be useful in noisy environments.

smooth

This parameter controls the smoothing of the noise reduction mask, impacting the transition between noise and signal. It ranges from 5 to 7, with a default of 1. Smoothing helps in reducing artifacts in the processed audio, providing a more natural sound.

seed

The seed parameter is used for initializing random number generators, ensuring reproducibility of the audio processing results. It accepts any integer value from 0 to 0xFFFFFFFFFFFFFFFF, with a default of 0. This is particularly useful for consistent results across multiple runs.

MW Audio Recorder Output Parameters:

audio

The output parameter audio provides the processed audio data in a dictionary format containing the waveform and sample rate. The waveform is a cleaned and normalized version of the recorded audio, ready for use in AI projects. This output is crucial for ensuring high-quality audio integration, free from unwanted noise and artifacts.

MW Audio Recorder Usage Tips:

• To achieve the best audio quality, choose a higher sample_rate like "48000" if file size is not a concern, as it provides better fidelity.
• Adjust the sensitivity parameter based on the noise level of your recording environment; higher values are suitable for noisier settings.

MW Audio Recorder Common Errors and Solutions:

Recording/processing failed: <error_message>

• Explanation: This error occurs when there is an issue during the recording or processing phase, possibly due to incorrect parameter settings or hardware issues.
• Solution: Ensure that your microphone is properly connected and functioning. Double-check the input parameters for any incorrect values, and try reducing the n_fft value if computational resources are limited.