Podcast Hosting Using Spectral Gating And Speech Recognition Methodology

Abstract

Podcasts contain information in an audio form which is being recorded in human voice. These recorded podcasts are later published on various podcast hosting websites where people can access and listen to various such podcasts. Podcasts can also be used to host an interview at a common location. But due to pandemic, the interviews are conducted through online platforms. The podcasts recorded via such platforms result in poor quality of the audio along with communication problems. To solve this problem, the paper focuses on building an application where interview-based podcasts can be conducted at different geographical locations while preserving the audio quality. Noise in the audio file is removed using Spectral Gating. Subtitles of audio file will also be generated using Speech Recognition algorithm. The final audio file will also generate an RSS feed link which can be used while publishing the podcast to notify subscribed users about the new updates. We have carried out our experiment on more than 100 audio files.