Let’s say you’re starting a podcast, or you’ve recorded an interview, or yourself speaking in front of the camera… And when reviewing the content, you realize the audio isn’t as good as you thought. To achieve slightly more professional results (once you already have your recording), you can use programs that simply remove background noise from your audio so the voice sounds much clearer.
Obviously, everything will depend on the original recording quality, but for cases where there is background chatter (like in a café or on the street), echo or other interferences such as the typical electronic hiss, the solutions we present below may be very useful to you.
Tips to remove background noise from your audio
With artificial intelligence
Recently, Adobe has launched its new system Adobe Podcast (inherited from Project Shasta) which, in short, uses its neural system to clean the audios we upload. All of this for free.
Simply, we have to drag our audio clip in MP3 or WAV format (if you have it in another format or in a video, you can use the free program Pazera Free Audio Extractor to convert or extract it), as long as it does not exceed 1h in duration and/or 1GB in size, and its AI will start doing the magic. At the time of writing, the service is in testing and has no usage limit.
Below you can listen to an example.
Original audio:
Processed by Adobe Audition:
The only drawback is that this artificial intelligence is trained on English audio, so in moments where some voices are incomplete or mumbled, the AI may “correct” the message by adding a generated voice in English. But as long as the main voice is loud, the background noise cleaning is more than adequate.
That said, another detail to keep in mind is that this portal will return the results in mono, even if the original file is in stereo.
Separate vocal and music tracks in songs with AI
Another resource that has been gaining traction lately are artificial intelligences capable of separating the vocal track of a song simply by receiving an .mp3, .wav or similar file.
So, if what you want is to separate the tracks of a song to obtain the instrumental and the vocals independently, you can use vocali.se for free while it is in beta. In this case it will only return the instrumental track and the isolated vocal track, rather than the different instruments separated into their respective takes, as services like LALA.AI do (one of the best in this field), although that platform does not allow you to download the results in its free model, only to play them on its website.
This last mention also has an option to remove background noise from your audios, and it even allows you to directly upload a video file and it will analyze only the audio itself.
.
