April 7, 2016 at 8:07AM

14

Recording better audio

I have been asked to record interviews for my school and I need help with the audio.
Ive done a video like this before but I wasn't exprienced and I tried my best to make the audio sound the best. Here it is
https://www.youtube.com/watch?v=V1wQGgFHVps

I used a Rode VideoMic Go and had my camera setting on manual, and at a -12 db.
In Adobe Premiere I used 3 audio effects:
DeNoiser: reduction -17.0 db and offset 10 db
Highpass: cutoff around 200 hz
Dynamics: presets - soft compression. selected compressor and auto. Threshold -22, ratio 4, makeup 12 db, attack 1ms, release 100 ms.
I found these settings that talked how to get rid of fuzzy background noise.

I can't buy a new microphone or another editing tool on the computer because I'm on a budget.

Im just asking if there is a way to get better sounding audio with my mic or in post-production?

15 Comments

Which camera are you using?

April 7, 2016 at 9:18AM

0
Reply
Guy McLoughlin
Video Producer
32544

nikon d3300. I don't know if that would affect the audio quality since I'm using an on-camera mic

April 7, 2016 at 10:42AM

0
Reply

The problem with the Røde video mic is that it tricks you into thinking it's OK to mount it on your camera. Generally speaking, that's the worst place to put it. Study this video and you can get better sound without needing to buy a better microphone:

http://nofilmschool.com/2015/08/24-sound-basics-will-help-you-lay-sturdy...

(In fact, spending more money on a microphone without understanding these principles is just a waste of money.)

P.S. If you find this answer helpful, don't just say thanks. Actually upvote the answer so that others can know that it was helpful.

April 7, 2016 at 9:55AM, Edited April 7, 9:56AM

1
Reply

Thanks! (didn't know about the upvote)

April 7, 2016 at 10:44AM

0
Reply

Your cutoff of 200Hz is too high. I understand why want to do it but the only good solution is to put the microphone closer to the speaker.

Next time get a mic stand with arm and position the mic above and slightly before the speaker off the camera frame or alternatively use a lavalier microphone.

A small piece of feedback about the video, please consider if putting people 'against the wall' is the best way to do an interview. I would suggest you try to create some space between the interviewee and the background.

April 7, 2016 at 12:00PM, Edited April 7, 12:31PM

3
Reply
avatar
Cary Knoop
Member
2291

My absolute best answer is to get the audio 'correct' when shooting, then you'll only need VERY few adjustments in post.

Use this little guy, it's $25, has an 18 foot cable. Anyone can afford this one to get better audio, it'll get you started:

http://amzn.to/1XjLFup

The closer your mic is to your subject, the better. A shotgun mounted on your camera for interviews unfortunately will VERY rarely work out well...

April 7, 2016 at 1:48PM

0
Reply
avatar
Jordan Mederich
Documentarian / Filmmaker
1368

Some basic rules of "thumb" when recording audio...

1- Never place a microphone more than 3 feet from the person speaking, which means you either have to boom your microphone ( RODE VideoMic + 10 foot RODE 3.5mm mini-plug mic extension + some sort of boom pole which can even be a broom handle ) or switch to using a LAV mic.

2- For most DSLR / Mirror-less cameras use a recording level of -6 dB instead of -12 dB because these cameras have a high noise-floor and you will lose a bit of your audio quality by using a -12 dB recording level. There are exceptions to this rule such as the Panasonic GH4 camera which has a low noise-floor of -72 dB, so using -12 dB with the GH4 you are guaranteed to have at least 60 dB of audio dynamic range in your recording.

3- When processing your audio follow these steps ( these steps are for beginners, experienced audio people will have their own audio post workflow process )

- DO NOT EVER use any type of "Normalize" function in your audio editor or video editor. Every "Normalize" function I've come across does terrible damage to the quality of your audio recording.

- First convert all of your audio tracks to 24-bit because the 24-bit format will help to preserve the quality of your original recording, where 16-bit audio can cause distortion when processing. If you are using an external audio recorder then always record your audio using 48 kHz 24-bit format. ( most low cost recorders will support this format )

- Next adjust the audio levels of your recording using your editor's volume control. Do this by selecting sections of your audio that are too loud or too quiet and adjust these sections individually. You want to end up with a finished audio that has audio PEAKS roughly in the range of -6 dB to -12 dB. Do not try and do this by selecting the entire audio track unless your entire audio track shows the same recording level in it's wave-form pattern.

- Next use your audio editor's EQ controls ( I prefer to use paragraphic EQ ) to improve the quality of your recording. You may want to boost or cut the low frequencies, mid frequencies, or high frequencies. You also may find that cutting the 800 Hz frequency will help to improve spoken dialog. ( test by trial and error to figure out how much of the 800 Hz frequency to cut )

- Next, if the dynamic range of your recording ( the difference between the loudest sections and the quietest sections ) is too high then you may want to apply some compression to your audio track to reduce the dynamic range. With many editors applying compression will lower the volume of the loud sections of your recording, so you will need to compensate for this by bringing up the volume of the entire audio track. I prefer to apply compression only to the loudest sections of my recording, and I adjust my audio compressor to effect only these sections. Again you are aiming to have finished audio peaks between -6 dB to -12 dB.

- Next apply noise-reduction to help remove any hiss that might be in your recording. Do this in multiple passes because applying too much noise reduction at once will destroy the audio quality of your recording. ( always check to make sure that the noise-reduction is not effecting the quality of your recording )

- All of these steps should be applied to separate individual audio tracks, so you are NOT trying to apply this to separate dialog tracks, sound FX tracks, music tracks at the same time. Everything should be processed separately.

- Lastly adjust the volume of your separate audio tracks to mix them together to create your finished audio track. Be careful with how you use music as it can compete with your dialog audio and make it difficult to clearly hear what people are saying.

April 7, 2016 at 2:39PM

4
Reply
Guy McLoughlin
Video Producer
32544

You wrote: "First convert all of your audio tracks to 24-bit because the 24-bit format will help to preserve the quality of your original recording, where 16-bit audio can cause distortion when processing. "

Not sure why you think that is true as quality audio editors use 32bit floating point representation of the audio.

Also I am not sure how you conclude that normalization causes terrible damage to your audio.

What audio editor do you use?

April 7, 2016 at 9:40PM, Edited April 7, 9:42PM

31
Reply
avatar
Cary Knoop
Member
2291

I started saving all of Guy's audio advice to my hard-drive.

April 9, 2016 at 7:48PM

4
Reply
avatar
Alex Zakrividoroga
Director
3759

You're not alone.

April 10, 2016 at 10:26AM

5
Reply
José Pedro Pinto
Wannabe
753

>>>Not sure why you think that is true as quality audio editors use 32bit floating point representation of the audio.

I've been using Sony Sound Forge Pro and a few other audio apps since the late 90's when it was owned by it's original creator, Sonic Foundry. In testing some extreme audio manipulation I have found that you will get a better results if you first convert all your audio files to 24-bit. Despite your DAW using a 32-bit work space some effects may be applied using 16-bit processing if the original file is 16-bit, so converting everything to 24-bit guarantees that no 16-bit processing will take place.

>>>I am not sure how you conclude that normalization causes terrible damage to your audio.

The problem stems from the normalization function applying a uniform maximum amount of gain to your entire audio track, when the audio track gain should be applied selectively section by section. Also with some audio editors normalization is performed at 16 bits using truncation, despite the DAW being able to handle 32 bit floating math, so your audio may pick up subtle distortion just by normalizing your tracks.

If you check out a few professional audio recording forums you will see that many pros won't touch any form of normalization with a 10 foot pole because of the damage it can do to your audio tracks.

Here's something to check out in terms of using normalization with your audio tracks...

10 Myths About Normalization
http://www.hometracked.com/2008/04/20/10-myths-about-normalization/

April 8, 2016 at 1:17AM, Edited April 8, 1:49AM

1
Reply
Guy McLoughlin
Video Producer
32544

You wrote: "The problem stems from the normalization function applying a uniform maximum amount of gain to your entire audio track, when the audio track gain should be applied selectively section by section."

So when you apply gain to a section only it suddenly does not "cause terrible damage" to your audio?

Sorry but I am not convinced, normalizing is just fine and if you use a DAW that does 16 bit processing I suggest you get a better DAW.

What worries me more is repeatedly apply noise reduction as you suggest you do, that definitely can harm the audio.

April 8, 2016 at 10:21AM

38
Reply
avatar
Cary Knoop
Member
2291

All you really need to do is use a lav mike or get the mike close to the person speaking and keep out extraneous background noise (turn off air conditioner, etc.). Set your volume to -6 to -12 and you should be good to go.

April 9, 2016 at 11:07AM

0
Reply

>>>So when you apply gain to a section only it suddenly does not "cause terrible damage" to your audio?

When you apply gain selectively you are essentially evening out the recording level of that track. It's a rare event where I see a complete audio track with one recording level throughout the recording. Yes, you see this when recording ADR or voice-over dialog in a studio, but most of the time the actual recording level goes up or down depending on the position of the mic and the performance of the person speaking. Modifying gain to selective sections of an audio track will alter the noise-level for the sections you modify, so ideally you should do the NR work first but with a good recording this noise difference is pretty minor.

With beginner's I often see people using normalization like a sledge-hammer when trying to improve the quality of their audio, when they should be figuring out what sections of their recording need more or less gain to even out the finished recording level of that track.

>>>if you use a DAW that does 16 bit processing I suggest you get a better DAW.

Unless you test all functions of your DAW you may not know that 16-bit processing is taking place on 16-bit files. I have run into this with Sony Sound Forge Pro in the past, and the cure was simply to re-save all 16-bit recordings in 24-bit format. ( a very simple task with most audio editors )

>>>What worries me more is repeatedly apply noise reduction as you suggest you do, that definitely can harm the audio.

Actually it's the opposite. If you test adding the SAME amount of NR in four passes compared to applying the SAME NR in one pass, you can hear that the multi-pass route produces a better result with less distortion or artifacting.

April 9, 2016 at 1:09PM, Edited April 9, 1:14PM

5
Reply
Guy McLoughlin
Video Producer
32544

The new Youtube Link https://www.youtube.com/watch?v=deo1a0Uk4U4
I had to take that video down and replace it, I accidentally spelt copyright wrong at the beginning...

May 3, 2016 at 10:01AM, Edited May 3, 10:01AM

0
Reply

Your Comment