Audio Spiking in video - How can I reduce/remove this?

[twiggy0]

Hello MacRumors community!

I am an avid videographer. I recently filmed a music video for a good friend of mine with a great voice using a 5D II and a rode video mic.

The problem I have is that the audio recording gets somewhat distorted when she is singing the chorus, while her voices pitches high. Sort of like a screeching static noise.

Would anyone know a program I could use to edit this to reduce or hopefully remove these spikes? I'm editing it in Final Cut 10, and also would be able to use audacity to edit the sound file if that could do the trick.

Any help would be much appreciated!

EDIT: Here is the link containing a sample of the music video. It's 1 minute long. The last 10 seconds have noticeable spikes in audio.
http://www.youtube.com/watch?v=QBx5KwTsfVE&feature=youtu.be (watch in 1080p for best audio quality)

While I'm at it, any criticism on the actual video would be welcomed with open arms. How could I have improved it? (Apart from the spikes obviously )

 

[simsaladimbamba]

Did you record on an external audio recorder? Is the recording analog or digital? If it is the latter, and the levels peak too much, there might not be much you can adjust, except some audio engineer magic I have only seen once or twice, but can't explain, as audio is not my main field.

And if you want to upload a sample, YouTube might not be the best, unless the audio is already included in the video source file. If it is an extra .wav file you have, if you recorded externally, then a sample of that .wav file would be better.

 

[twiggy0]

Did you record on an external audio recorder? Is the recording analog or digital? If it is the latter, and the levels peak too much, there might not be much you can adjust, except some audio engineer magic I have only seen once or twice, but can't explain, as audio is not my main field.

And if you want to upload a sample, YouTube might not be the best, unless the audio is already included in the video source file. If it is an extra .wav file you have, if you recorded externally, then a sample of that .wav file would be better.

I recorded with this microphone:
http://www.amazon.com/Rode-VideoMic...9SOC/ref=sr_1_1?ie=UTF8&qid=1334126305&sr=8-1

that was attached directly to the 5D Mark II. So the audio is basically attached to the video.

I believe it would be considered digital audio. So I can't do anything to remove them? It takes away so much from the video.

Thank you for such a fast response by the way.

 

[simsaladimbamba]

That is an okay microphone for its price range.
Maybe have a sample for us, to see, what can be done.
Do you edit the video in ProRes or do you edit the footage natively?
If it is the former, maybe export a sample using that codec (5 to 10 seconds) and upload that, or if you can just export the audio part*, that would be great.

Also know, that it is better to use a dedicated recording device for audio, where you can set the levels manually and have a better limiter to avoid such peaks. Sometimes a mixer might do too.

* FCP X > Share > Export Media (CMD+E) > Audio Only > WAVE or AIFF.

 

[twiggy0]

That is an okay microphone for its price range.
Maybe have a sample for us, to see, what can be done.
Do you edit the video in ProRes or do you edit the footage natively?
If it is the former, maybe export a sample using that codec (5 to 10 seconds) and upload that, or if you can just export the audio part*, that would be great.

Also know, that it is better to use a dedicated recording device for audio, where you can set the levels manually and have a better limiter to avoid such peaks. Sometimes a mixer might do too.

* FCP X > Share > Export Media (CMD+E) > Audio Only > WAVE or AIFF.

Here is a link to the sample video: http://www.youtube.com/watch?v=QBx5KwTsfVE&feature=youtu.be
Noticeable spikes can be heard at the last 10 seconds. Also while I'm at it, any criticism would be welcomed with open arms.

I'm editing everything using Final Cut 10, not using any other program. What external microphone would you recommend me to use in the $200-$300 price range?

I also own this microphone: http://www.amazon.com/Rode-NT1A-Ann...cal-instruments&ie=UTF8&qid=1334127997&sr=1-1
and was hoping to get a Tascam DR100 to be able to provide it phantom power. Would that be a great sound setup? I would have to buy the Tascam.

 

[floh]

I also fear for the worst... The 5DII, while delivering amazing video quality, lacks some audio adjustments that would make your life easier (and they have been added to the brand new Mark III). This peaking probably happens when your sound input is amplified too loud before being recorded digitally.

A digital audio recording (just like an analog one...) has a maximum amplitude that can be stored. The "x-axis" and "y-axis" if you look at your audio clip are divided into many small steps and at each time step (x-axis) an audio value (y-axis) is stored. That's put very simply, but not far from the truth. You can imagine that there is a lowest noise you can record (one step up on the y-axis) and a loudest noise (all steps up to the very top).

The trick when recording audio is to set the pre-amplification of your microphone in such a way that you want as much signal as possible, while just not reaching the highest value. If the pregain is set too low, you will hear lots of hissing and noise if you amplify it in post. If it is set too high, the audio will peak and sound distorted. I'm guessing that's what happened to you.

If this happens once or twice with an actor shouting or clapping, you can save it in post production with the mentioned "audio magic"... But if a singing voice peaks, you will not be able to make that sound good afterwards. There is just information missing above the maximum sound level.

I'm pretty sure (I don't own one) that the 5DII has a setting for the audio level, where you can turn down the pregain. You should try to do that. But the recording you have now is probably useless.

The Tascam DR100 is a nice tool which I also considered buying at one point. It will have the advantage of showing you the recording level at all times so you can adjust the pregain smoothly. And yes, it has phantom power. Also take a look at the Zoom H4n. It is always recommended as an external recorder and that is for a reason: It delivers the best bang for the buck. Not as tough in built quality as the Tascam, but a little better with the internal microphones.

 

[twiggy0]

Thank you floh for such a detailed response!

If I get the DR100, would it be worth attaching my rode NT1A diaphragm condenser mic and recording on the NT1A, or just throw that completely aside and only use the DR100 for music recording?

I filmed it with CP.2 lenses and a $150 microphone. I really did skimp out on the microphone.

I'll be sure to look into both those model microphones for my next music videos.

 

[simsaladimbamba]

The last times I recorded audio on set I used ME64 and MKH 416 microphones from Sennheiser. Since they are directional microphones, they might not be what you want, though it would be nice to have one microphone for her voice and microphone for the piano.
And as far as I can tell from the video, the audio is not usable.

 

[twiggy0]

The last times I recorded audio on set I used ME64 and MKH 416 microphones from Sennheiser. Since they are directional microphones, they might not be what you want, though it would be nice to have one microphone for her voice and microphone for the piano.
And as far as I can tell from the video, the audio is not usable.
And why is it in 5.1?
Normally recorded sound is mono if you only have one microphone.

The audio is not usable because of the spikes, or it just generally isn't good?
I have no idea why it's in 5.1, I haven't edited any settings.... I would imagine it to output it in mono automatically. I'm not really savvy in terms of editing the audio on my 5DII, thought I know all the in's and out's of the video part.

Could I do this.... Tascam DR-100 pointing at the piano, with the NT1A condenser mic I already own recording her voice (that's attached to the tascam to give phantom power) Would that work? (Also if this would work, I would have absolutely no idea how to work with the files. Would there simply be two separate audio files recorded onto the tascam (one being the tascam, the other being the NT1A) which I could then combine in final cut?)

Also, would the NT1A I own be able to record the loud spikes when she sings the chorus without problem? (Ironically, I own that microphone but never got the chance to use it because I don't have anything that provides phantom power)

 

[simsaladimbamba]

The 5.1 sound was an error on my part. Forget it, it's a mystery.

I would use that Tascam and use both its XLR inputs to connect your two microphones to it to record singer and instrument as good as you can. As the Tascam probably has a good limiter (my DAT recorder from Tascam has), you will also not have that much problems with peaks and can concentrate on the video part.
Just make sure to record audio with the 5D too, to make synching easier.

 

[floh]

Okay, I'm back from giving a lecture to give a lecture...

No, just some short responses: Your Rode NT1A is a fine mic (I've used it before) which will sound way better than the DR100's internal mics. Those are very suitable for ambient recording to later add to your videos, but not for recording a voice up close. The NT1A might even look cool in the video when placed in front of your singer. It would definitely be a good idea to use this mic in the recording.

Unfortunately, I can not recommend you one mic in the range 200-300 euros that will be suitable for everything. Microphones, even more than lenses, are highly specific to be used for a certain purpose. Your condenser mic is very nice for close-up vocal recording but will be of no help in recording the room sound of a wide-angle scene, since it will be mumbled. The shotgun (directional) mics that simsaladimbamba mentioned are very good for recording actors' speech from a little distance (=from outside the image), but you would not want to record any music with them.

I would also recommend you to use both mics, the Rode NT1A for the piano and the Videomic pointed at the piano. That way, you can even mix it later if either piano or voice are too loud.
By the way: This is not an electric piano, is it? Otherwise just take that signal into the Tascam, too! Recording a piano sound at good quality is a science on its own and people use around 3-7 microphones just for the piano. You will never get as good as the samples in an electric piano when recording, sadly.

The bad news is: As far as I know, the Tascam and most other recorders can only record either their internal microphones or the external ones on the XLR plugs. So your idea of pointing the Tascam at the piano and adding the NT1A will most probably fail.

I start rambling again. Good luck with the recording.

 

[Zwhaler]

Is it just me or am I not hearing very noticeable spikes in the audio? For the last 30 seconds I heard one or two minor peaks in her voice... If the last 10 seconds concerns you simply blade or split the clip right where the increase in levels starts, and lower that volume by -1dB or more until you are happy with it. You can further split it to make more detailed adjustments throughout the last 10 seconds.

 

[simsaladimbamba]

Is it just me or am I not hearing very noticeable spikes in the audio?

You can hear distortions from 00:42 (mm:ss) onwards. Such audio, unless news maybe, would never get on the air, with exceptions of course. It is very noticeable for me and distracts from the whole song.

 

[wonderspark]

I've rescued peaked audio with painstaking tiny samples of good audio pasted over the bad, but that is not fun.

I also have a Tascam DR-100, and the limiter on it is amazing. It's my first choice for all my mobile sound recording right now. I use the XLR inputs: one for a Sony ECM-77b lav mic, and the other for a selection of Sennheiser shotguns on a C-stand or boom pole operator.

I've synced sound so much that it's way easier for me to just record it separate and sync later. Every clip I've ever synced that also had sound from the onboard camera mic has been at least one or two frames off anyway, so I prefer a dedicated recorder for more accurate sound as well. (The farther the camera is from the sound source, the further off the sound will be.) I use the camera mic waveform to help guide the match-up of my dedicated sound file, then delete the camera audio.

 

[Zwhaler]

You can hear distortions from 00:42 (mm:ss) onwards. Such audio, unless news maybe, would never get on the air, with exceptions of course. It is very noticeable for me and distracts from the whole song.

Okay that helps because the OP said the last 10 seconds so I wasn't even going back that far.

 

[monokakata]

Twiggy0 --

You have Audacity, so you can check the audio to see if fixing it's hopeless or not. Bring up the audio in Audacity and inspect the waveform.

If the bad parts are flat (=horizontal) at the peak, then that's hard clipping and there's nothing practical for you to do (except hire wonderspark).

I've attached an image of clipping, so you can see what it looks like. There are no more bits left to use. Decreasing the level can't remove the distortion.

 

[twiggy0]

Thank you for everyone's help.

I'm going to end up having to deal with these spikes for now and be looking to purchase the DR100 from tascam to attach my NT1A mic too for future videos.

As always, you guys have been a great help!

Thanks again.

If anyone has any advice on how I could improve the actual filming of the video, that would be welcomed!

 

[floh]

If anyone has any advice on how I could improve the actual filming of the video, that would be welcomed!

I'm not an expert, but I can give my two cents if you like me to...

Some general hints:
- If your camera isn't moving, don't go handheld but fix it on a tripod (they are way cheaper than the Tascam recorder). The very, very slight shake isn't necessary. Especially since "shakiness" isn't really what the song's message is about.
- You were filming from two positions, but both were at the exact same crop of the image (close-up). Maybe you can vary that and show the piano. I mean, the girl is playing herself, that's impressive... Or maybe add a detail shot on her mouth when she sings very soft parts. Just spice it up.

If you want to go more into the direction of a "music video" that might be seen on TV, here are some more ideas:
- Usually, the audio is recorded in one take (normally in studio) and then mixed and produced. For every following take, the original audio track is played back and the artist can either sing along or pretend to. That way you can get fancier stuff and multiple camera angles even if you only own one camera.
- You can add some movement to the shots. You don't need a dolly or things like that. Just an example: When she goes into the refrain, you can pull away by having your tripod stand on two legs and just tilting it away. You would have to follow focus though (from the looks of the video, I assume you are filming DSLR...). The effect would be an emphasis on the song's change from intimate storytelling to "smile, sunrise and everything opens up"... put a little flowery...
- If you still want to do more, you can just watch some MTV music videos and get inspired of whatever crazy stuff you can find there.

I actually liked the lighting (from a window on her right?). Getting something better would be a lot of effort.

 

[twiggy0]

I'm not an expert, but I can give my two cents if you like me to...

The lighting was actually done with a studio kit I bought and all ambient lights were turned off, which is why the contrast appears so clearly. Very simply but I decided to do it that way once I realized that turning off the front light creates the beautiful contrast.

I did use a DSLR. I got multiple angle, in which some did move but wasn't really good since I was trying to imitate a slider motion by walking sideways... I didn't include it in this because it's a sample. Which tells me I probably should upload the final one before getting more criticism lol.

The idea of the zoomed in lips is awesome though, as well as mounting the tripod and pulling it back on two legs while keeping her in focus, I've seen a youtube tutorial on this, will definitely try that on a refrain!

 

[musique]

OK, here's another unsolicited two cents

We've become so used to dolly shots on TV shows that it's almost rare to find a camera not moving for a longer clip (more than, say 8 or 9 seconds). Although your subject is interesting to see, if you are able to reshoot, you might consider a few other angles. For one, there is always a nice shot on piano keyboards with the polished wood and the player's hands. Depending on the room (it seems there are white walls and a set of stairs behind her) you might shoot her from the rear, perhaps 3/4, and also if you can do it, from up the stairs. These would be mainly for a change from the focus on her face. (Sort of a redefinition of "B-roll.")

One last suggestion: Is it possible to add some very soft sweetener like a string ensemble (synthesized, unless there's a budget) and possibly a couple of French horns to "glue" the group together? I've done this in things that I've scored under a solo piano and it's added to the success of the music.

I like the look of what you've done. Good luck.

 

[SchwartzSound]

- Usually, the audio is recorded in one take (normally in studio) and then mixed and produced. For every following take, the original audio track is played back and the artist can either sing along or pretend to. That way you can get fancier stuff and multiple camera angles even if you only own one camera.

+1 This is the way to do it. Not only will it free up your camera to do more fancy shots, but the audio itself will be better off for it. This way you can get ideal mic placement and without having to worry about aesthetics, and record until you get the best music take.

I would even recommend recording the piano completely separate to the voice. It sounds like she's playing an electric keyboard, so you could still record simultaneously -- simply take the direct line outs of the keyboard (no live sound) and record them while you mic her voice, and have her monitor the piano on headphones while singing. This way you'll be able to mix and process the piano and voice individually later on if you want to.

 

[twiggy0]

We've become so used to dolly shots on TV shows that it's almost rare to find a camera not moving for a longer clip (more than, say 8 or 9 seconds). Although your subject is interesting to see, if you are able to reshoot, you might consider a few other angles. For one, there is always a nice shot on piano keyboards with the polished wood and the player's hands. Depending on the room (it seems there are white walls and a set of stairs behind her) you might shoot her from the rear, perhaps 3/4, and also if you can do it, from up the stairs. These would be mainly for a change from the focus on her face. (Sort of a redefinition of "B-roll.")

One last suggestion: Is it possible to add some very soft sweetener like a string ensemble (synthesized, unless there's a budget) and possibly a couple of French horns to "glue" the group together? I've done this in things that I've scored under a solo piano and it's added to the success of the music.

I like the look of what you've done. Good luck.

Well it was for a friend actually, so the budget was $0, and I've got no idea how to use Logic Pro. I tried a shot on the stairs but it ended up not working, even with a 85mm lens. It just seemed too wide and her back too boring. I'll try the dolly shots more, I finally got my heads on a steadicam so that should help. I like how you're suggesting to take more different shots instead of the same subject from different angles.

Thanks!

 

[treatment]

A couple of things:

The trick when recording audio is to set the pre-amplification of your microphone in such a way that you want as much signal as possible, while just not reaching the highest value. If the pregain is set too low, you will hear lots of hissing and noise if you amplify it in post. If it is set too high, the audio will peak and sound distorted. I'm guessing that's what happened to you.

Although this was true in the analog domain, it is simply not the case anymore.(in the digital domain) The A/D convertors in any digital device will add noise to the signal. Higher quality DAC's = less noise. Camera's do not offer people high quality DAC's. EVER. High quality DAC's on a camera is simply a lower manufacturing priority than all of the other things a camera NEEDS to have.

Although distortion of the signal is what we are trying to avoid, keeping the mic preamps LOW, is a much better idea, to accommodate peaks, and then you can boost it up later. (Normalize/Comp Limiter/EQ...the basic tools)

You are going to get noise regardless, and that noise is going to be the same level in ratio with the level of the signal. In short, it doesn't matter whether you record it too low, the priority is that you don't record it too high.
(I hope I am being clear enough about this, without getting into signal-to-noise ratio arguments, and discussing headroom)

Recording a piano sound at good quality is a science on its own and people use around 3-7 microphones just for the piano. You will never get as good as the samples in an electric piano when recording, sadly.

Although true, there are workarounds that haven't been mentioned here:
Although most electronic pianos "out of the box" tend to be "okay" for live performance, and sub par for recording, there is always the solution of taking the MIDI out, and into a software instrument.
As you said, mic'ing a piano is a skill. And Lord knows I've done it countless times. There have been times when it worked wonderfully, and times where the phasing was so bad, the piano sounded out of tune, or the left to right balance was out to lunch.
The software for pianos in the last 5 years has grown so much that the sound quality of the samples is typically BETTER than what your average person could accomplish in a home studio through trial and error.

On mic choice/placement:
There are a bazillion ways to skin the cat, I would rather not get into it, but more importantly:

When making a music video, it is VERY uncommon for the performance to be the actual performance. It is just not logical.

Record, mix, and master the recording, and do the video later.
This eliminates your mic questions, because they are no longer relevant. The artist is lip synching to her own music, and that is what professional artists EXPECT to do.

On the video itself:
I really like the color tones in the video, although I have definitely seen these kinds of colors before, and in quite a few recent videos.
The depth of focus is also really nice.

I would like to see a shot of the room she is in...maybe a wide angle profile shot. An over the shoulder shot of her fingers on the keyboard, a close up shot of the fingers on the keyboard. We don't really get a feeling of "space" in this video. Even if the room she is in is a cluttered/ugly bedroom, you could shoot the wide shots elsewhere in a different location.

Sometimes the music video is like a movie, and tells a story. Sometimes, it is about the artist and the performance of the song. It appears that you are trying to accomplish the latter, so I would suggest the above just to add some depth to it.
Hope this is helpful.
Treatment

 

[linuxcooldude]

Since you are using Final Cut Pro 10 they have several compressor plugins that will take the spikes out for you.

 

[logicpro7]

I havent read all the posts, so advanced apologies but does the singer have a studio recording of the song? As in has she recorded this in a studio?

If so sync the master up with the video and delete the audio from the video files.

I would suggest she goes into a studio and records it if not.