Audio Troubleshooting and solutions in Audacity
If you have suggestions on how to improve this document, or find mistakes, please send them to firstname.lastname@example.org
This is a troubleshooting guide for problems you might encounter with audio recordings.
The solutions make use of the audio editting program Audacity
- The audio is not loud enough
- Loudness in seperate recordings varies too much
- Within one audio clip, sound levels fluctuate too much
- There is too much noise in the audio recording
- The recording sounds metallic or unnatural
Audio levels are measured in dB’s. For a (digital) recording, the highest peaks should reach 0 dB but not exceed it. In an optimal recording, lower peaks come close to 0 dB as well, but don’t reach it.
If your audio recording is not loud enough, select the entire track in audacity and then click Effect –> Normalize
Make sure “Remove any DC offset (center on 0.0 vertically)” and “Normalize maximum amplitude to:” are selected. Also check that the white textbox has the value -1.0 dB
If your recording was a stereo recording, select “Normalize stereo channels independently” for optimal effect, but stay clear of this option if you are not sure.
Hit OK when you’re done. The audio should now be leveled to -1.0 dB
If it is still not loud enough, undo normalization, try the “Hard limiter” option and then normalize again.
The compression filter in Audacity will increase small peaks and decrease low peaks in your file, to make the loudness more continues, while retaining some loudness variation.
To use compression, select the entire track that you wish to compress, or a part of that. Go to Effect –> Compressor. You will probably be fine leaving the settings as they are, but if you know what you’re doing, you can play around with them. Hit OK when you’re done.
The sound should now have a more continues loudness. The loudness of the file has been changed, but all other parts of the sound remain unchanged.
If this did not fix your problems, there are probably a few peaks that are way larger than the others.
The height of the peaks indicate the loudness of the sound registered at the point in time that peak corresponds to.
If a lot of your recording has the same loudness (meaning most of your peaks reach about the same height, let’s call those peaks the normal peaks), with a few exceptions (let’s call those peaks the exception peaks), you should use the Hard limiter. Those peaks most likely correspond to unwanted sounds like someone coughing, touching the microphone, a door slamming, etc.
First of all, find the loudness in dB of the peaks that you don’t want to remove. You can do this by clicking the triangle next to the name of the file and selecting “Waveform (dB)”.
You will see a scale from -60 to 0 to the left of the waveform. You can drag the bottom line of the waveform to enlarge it.
Find the dB level of one of the highest normal peaks.
Go to Effect –> Hard Limiter
Leave the Wet level at 1.0 and the residue level at 0.0 but change the dB limit to a little bit more than the number you just found (so if the highest of the normal peaks is -13.0 dB, use, for example -5.0).
Play around with this value so only the exceptional peaks that you want to remove get clipped, but none of the normal peaks are. You can either use the preview function, or hit OK and undo and retry it every time too much or too little gets clipped.
Caution is advised when using this feature. The Hard Limiter clips all peaks louder than a certain dB value. This means information gets lost. With most exceptional peaks, this information is not relevant, but if parts of speech get clipped, they will sound very ’80 computer-like.
Noise is a constant hiss or hum in the background of your recording. It can be annoying, and in bad cases can make recordings unusable. So always listen back to your recordings immediately and keep an ear out for it. It is always better to prevent noise from occurring in the recording than to have to remove it after the fact.
Poor quality or faulty equipment can introduce noise into your recording. For example, using your phone to record stimuli will probably result in a noisy recording no matter what other steps you take. Thankfully, you don’ t have to record with your phone, because the lab lends out all kinds of high quality audio equipment! When inquiring about borrowing some equipment, make sure you lay out your quality requirements to the lab support staff. We will recommend different equipment depending on your requirements.
Besides equipment, the most important factor in reducing noise is managing the signal-to-noise ratio. This involves, perhaps obviously, increasing the sound level of the signal and decreasing the level of the noise. To achieve this, try the following:
- Position the microphone further away from obvious sources of noise, such as air vents, loudspeakers, and windows.
- Use a wind cap on your microphone and position yourself accordingly to prevent wind noise when recording outside.
- If you’re not sure where the noise is coming from, move the microphone around while monitoring the sound level to find a position and direction that improves the situation. If this step has no effect at all, your noise source may be internal to your recording equipment.
- If you have multiple devices in the audio chain with adjustable volume, try to identify the device that introduces the most noise. For example if both your microphone and amplifier have controllable volume, you may get better results turning up the microphone volume whilst turning down the amplifier gain (or vice versa). If your noise source is external, then this step will have no effect or will reduce the signal level as much as the noise.
- Increase the sound level of the signal by moving the microphone closer to the source (i.e. the speaker’s mouth) as much as is possible without affecting the recording quality. This will in turn allow you to turn down the volume of the entire recording which, assuming that the level of noise remains constant, will reduce the amount of noise in your recording.
- Try unplugging and reconnecting all cables to identify a faulty connection.
NB: When recording in the phonetics lab, there should be no audible background noise at all. If you are nevertheless experiencing noise issues, please first verify all your settings match the how-to Recording audio in the phonetics lab. If the noise persists, please contact lab tech support as it may be a wiring or computer issue that must be addressed.
There is noise in your recording if you hear a (slight) hiss or hum in your recording.
This is not necessarily a problem, but if this makes your recording hard or annoying to listen to or even completely useless, you should probably try to remove it.
For a tutorial on noise removal in Audacity, watch this short youtube video
Make sure to select a silent part of your recording for the noise profile, otherwise too much detail will be lost and your recording will sound very unnatural after the process.
The human voice operates in a wide range of frequencies. If some of those frequencies are not recorded, the recording will sound unnatural.
To fix this, you can use Effect –> Equalization.
For instructions how to use this effect, watch this video
If there are topics that you would like to see in this troubleshooting guide, please contact Jan de Mooij.