Manipulating sound files (remove "temple background" sound)

Question

A friend of mine helps me to record Thai phrases in Thailand. I send him some phrases everyweek and due to bad resources in his village I have to wait a rather long time to get the recordings sent back. The first time I sent a few phrases to test if his voice and soundquality was OK. I was surprised. No clicks and pops and so on. But this time - the second time a got some phrases back - the sound is as before, no clicks, pops and so on - But, I fear he has been recording in a Buddhist temple or something, or even a toilet :-P. Would be wrong to say echo but you know how it would sound if one is recording in a temple or something.

Instead of contacting him again, I wonder if this "temple voice sound" somehow could be removed? Is there a good filter for this? I am using Audacity and trying the effects that comes with the program (high pass filtering, reverb and so on - but none of them works).

I hope that you understand my question and hopefully can provide an answer

Thanks!

Answer 1

There are a few things you can try but it's impossible to say without actually having the sound file in front of me. All of these suggestions will make you recordings sound messy and unprofessional but might help separate the voice from the reverb.

1. There are usually expander presets in dynamic processor FX. The dynamic processor might also be called a compressor, although the expander effect is actually the opposite of a compressor. The expander will take the loud sounds and make them louder, and it will take the quiet sounds and make them quieter. Assuming that the vocals are above the right thresh-hold, this will cut out some of the decay from the reverb.

2. Similar to the first suggestion you could achieve the same result with a noise-gate (although if you already can't hear what he is saying, this probably might not help)

3. You said that you tried a high-pass filter, but I'm pretty sure a temple would have more frequencies that you could remove to try to isolate the voice. Similar to finding the sweet spot in a recording, you would want to use an EQ or filter to tune into the vocal sweet spot of the voice while cutting out the extra frequencies of the reverb.

4. Channel extractor effect: this might only work if the recording is stereo. If it is a stereo recording you might be able to tune into the stereo position that contains more vocal and less reverb. This is more of a technique used to extract vocals from professional mixes so it might be harder in your instance.

5. Noise reduction (noise profile approach): Depending on your wave editing program, you might be able to select a part of the wave where there is mostly reverb and turn that into an FFT file, then use that file in the noise reduction effect process to "cut" that noise from your recording. I don't know if you can do that in Audacity, but I've done that in Audition https://helpx.adobe.com/audition/using/noise-reduction-restoration-effects.html

Good luck

Answer 2

Get him to do proper recordings.

There are some very advanced plug-ins for reverbation removal as well as background noise removal. But plug-ins never work in an exact way, you'll lose some information, the voice may start to sound worse.

http://www.zynaptiq.com/unveil/
http://www.izotope.com/products/audio/rx/

Answer 3

As Internet Human statet, this is a high end restoration job, that needs special Programms. But I want to add one little Plug-In to the list. It is the De-Verb plug in from Brainworxs that can, sometimes, work wonders to remove the sound of a room.

Answer 4

If getting him to re-record, consider asking him to use a dynamic mic such as the Shure SM-58 or SM-7b.

Being both dynamic mics with a cardioid pattern, they'll pick up considerably less room sound.

Answer 5

The best thing is, as others have said, to get it rerecorded in an environment that doesn't have a lot of reverb. However, if that isn't possible, you may be able to at least cut out some of the worst parts of the audio with careful use of a bandpass filter and noise gate. The quality will suffer, but if your friend is unwilling or unable to record in the environment that he did for the first set of sounds, that may be your only choice, aside from finding someone else who can record on a tighter timetable.