Automatic Vocal Pop Remover

Plosive sound in audio recording, especially in podcasts, can easily impair the listening experience of the audience. Wind blown into a microphone during “p” and “t” pronunciation causes plosive, also known as “pop” sounds. Many supplementary recording equipment like pop-filters and audio interface can be used to prevent those plosive audios. However, it is often expensive and painstaking for those just starting to produce podcasts. Once the plosive sound is recorded, it is hard to remove the sound in post-production process. To tackle this problem, an algorithm that can automatically eliminate the plosive sound without hindering the audio’s message was created. First, a sample plosive without pop-filters and model recordings with pop-filters were made for analysis of the pop. The main goal was coming up with a plosive elimination filter that will transform audio with the plosive sound into a model audio. This was done by statistically examining the equivalent audio with and without the plosives and by modifying the frequency contents and removing the plosives. The plosives mainly originated from lower frequency area around 200 to 400 Hertz. The algorithm surgically removed the plosives. The edited audio, compared to the model audio, sounded almost identical to the model audio, both audibly and visually, except for minor higher frequency ticks.