Written by , , , on

No matter how good your technical equipment might be, it is almost impossible to avoid capturing unwanted in-/exhaling sounds and mouth noises during voice recordings. After some users asked for an automatic removal of such sounds to improve the audio quality, we got to work and are now proud to present a major upgrade to our Denoiser, including the automatic removal of mouth noises and a new “Remove Breathings” option!
Check out our Audio Examples and the Getting Started Guide below.

What is new?

  • Remove Breathings: When the new “Remove Breathings” option is enabled, all the inhalation and exhalation sounds will be muted like all the other noises.
  • Eliminate Mouth Noises and Eating Sounds: With our upgrade to the Denoiser, we have integrated the elimination of all types of mouth noises, such as smacking, chewing and clicking. These sounds are set to silence in your output file.

Breath sounds are traditionally either manually silenced or removed via level-based gating algorithms using a pre-defined threshold to discriminate breaths from speech. However, depending on the speaker and recording setup, levels of speech and breath sounds may be very similar, making accurate breath removal difficult. If the threshold is set too low, some unwanted breaths may pass. Conversely, if the threshold is set too high, quiet speech may be unintentionally removed. In contrast to these traditional approaches, our new AI-based “Remove Breathings” algorithm is trained on a huge collection of breath sounds to automatically discriminate breaths from speech and to reliably remove the former while keeping the latter intact.

NOTE:
By default, mouth noises and breathing segments are muted but not cut. If you want to cut out all silent segments, just add the “Cut Silence” option for “Automatic Cutting” in the audio algorithms.

Audio Examples

1. Example: Breathing Removal

In our first audio example (Freak Show 136), a speaker breathes directly into the microphone, creating a Darth Vader-like effect. This heavy breathing is removed using the “Remove Breathings” algorithm and the Dynamic Denoiser:

Original:
Denoised:

2. Example: Mouth Noise and Eating Sounds Elimination

The following example is an excerpt from the Joe Rogan Experience #2054 where Elon Musk and Joe Rogan are eating pizza and talking about how terrible it is for the sound. Listen to how the chewing and smacking sounds are removed after processing with the new Dynamic Denoiser:

Original:
Denoised:

Getting Started Guide

For the Auphonic Web Service:

To remove breathing sounds in the Auphonic Web Service, simply create a production or preset as usual and check the “Remove Breathings” checkbox for “Noise Reduction” in the “Audio Algorithms” section.
“Remove Breathings” is only available for Speech Isolation and the Dynamic Denoiser, but NOT for the Static Denoiser.

For the removal of mouth noises, no special settings are required. Just enable the Speech Isolation or the Dynamic Denoise method for Noise Reduction.

For API Use:

To use “Remove Breathings” with the API, set the "denoisemethod" to dynamic or speech_isolation and the "debreath" flag to true, as in these two examples: 

"algorithms": {
    "denoisemethod": dynamic, "denoiseamount": 100, "debreathamount": 100,
}

"algorithms": {
    "denoisemethod": speech_isolation, "denoiseamount": 12, "debreathamount": 100,
}

For more information about the API, please visit our API help page, where you will also find all other API Audio Algorithm Settings.

Conclusion

Breath and mouth noise removal is essential for podcasters, as it enhances audio clarity and ensures a more professional, polished presentation. By removing distracting sounds, you can maintain listener engagement and provide a seamless listening experience.

Anyway, we hope you enjoy the new features!
If you have any feedback for us on how the breath and mouth noise elimination is working for you, please feel free to contact us via email or directly comment in our production interface!




Recent entries