Automatic audio processing

Automatic audio processing

Recently my client gave me one of his podcast recordings which he has processed through online audio processing service. While i had no claims about loudness levelling and normalization in this file, I noticed almost intact plosives (low frequency spikes — usually below 100 Hz, — that sound very unpleasant and annoying).

Here’s my little analysis.
Tool: iZotope RX6 (industry standard in Audio Restoration realm).

Working with loudness

Levelling and normalization — good job. On the second screenshot you can see the declining trend. In the second part of the file speaker has increased his intonations and subjectively sounded louder. So it was a correct job.


Picture 1 — Unprocessed sample waveform


Picture 2 — Processed sample waveform


Working with plosives

Almost ineffective. Developers offer you to use a High Pass Filter to reduce unwanted rumble and plosives, but actually, as you can see, it does not work as it should. I completely understand those guys… they, well, reinsured themselves and applied not so strong filter. Kinda compromise. But anyway, plosives were almost intact.


Picture 3 — Unprocessed sample (plosives)


Picture 4 — Processed sample (plosives)



Beyond any doubts, you can use online services for processing your home-made podcasts/interviews (and you should, because nobody wants to spend extra money), but if your aim is a more qualitative result, please use professionals‘ ears and experience and only work with those you wanna hug ツ

UPD: please check it out and see why you cannot always rely on automatic tools completely.

— Stan