Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Garynyc]

Hi all;

I am in a conundrum over a good workflow for editing conference recordings. We do the recordings live, on site at different conference centers, hotels and the like,  and  have to edit them with a one day turnaround. I have Sound Forge 10 and its a full  featured program, but I would like to streamline workflow and get a consistent result  over many different recordings. Sometimes the rooms are different, the mics are  different, the speakers styles are different, etc. One of the main issues I would like  to solve is the dreaded "panel discussion", where multiple speakers use their mics in  different ways--from the loudmouth to the low talker, the head turner and the off-mic  speaker unaware that there is a mic near them at all.

I would like to know if there is some kind of normalization method in Sound Forge where  I can actually have the program take all of these different speakers and put them  roughly on the same par with eachother. Currently I just click and drag over a low area  of speech and normalize to a level comprable to the others, then select the whole file  and bump it up if I need to. I simply cant take the time now that we have up to 15  simultaneous recordings being dumped to my PC at a time to edit all of them with lots  of care!

Other questions--approximately what decible range should the heart of the speech be at?  I am comfortable with peaks hitting between -3db--1db. Is that too conservative?

I use clipped peak restoration to take the edge off some recordings that are too "hot"  seems to work well. Is this an advisable plug in to use for overdriven speech?
If so would anyone recommend using the "batch converter" in order to automate these  processes? Or should each be taken on individually.

Off-mic questions. Is there a good method for getting them to be audible without making  them sound like a tin can?

I really help someone can help me out here. Thank you for any and all advice!

[lowland]

That's quite a can of worms to try and unravel, Gary, but here are my immediate thoughts:

Are the recordings being actively recorded with an experienced sound person present, or is it a set-and-forget arrangement where 'record' is pressed with the intention of sorting it out at the edit stage? There's no substitute for having someone proactively mixing in the scenario you describe, so I hope it's the former but fear it's the latter, as most of the problems you're talking about should have been reduced or solved with someone who knows what they're doing being there; also, get the material mixed at source and you'll considerably reduce edit time. Off-mic questions? - ideally there should be a roving soundie in the audience with a mic (shotgun type would work well) mixed into the main program.

Editing-wise, I'd think you need at least twice the raw run time to edit, could be quite a bit more depending on content and application. If you're the only editor and you have a one-day turnaround I'm not remotely surprised you're having overflow problems - you really need a team of editors on a job like this, each with their own edit machines and preferably a server.

It sounds as if you're new to this but have a great assignment there, which I hope will spin off into more work for a long time to come; however, it looks on the face of things that you haven't properly thought the requirements through, so if you're going to hang on to the commission you probably need to hire more people and buy more kit, or look forward to no sleep for some time to come. Either way, all the best with it: if you survive this it sounds like you'll emerge having learnt some important lessons on how to do that kind of work, a valuable knowledge-base for the future.

Hopefully you can see there's little in the way of a 'magic bullet' for a job such as this - it's down to preparation and costing sensibly so you a) have the resources you need to do things properly and avoid overworking yourself but b) end up making a profit so you'll want to do it again.

[Garynyc]

Thank you for your reply Nigel. Unfortunately, the conference conditions vary, from large ballrooms with a pro board operator to set and forget rooms that have 5-6 people in a 300 sq ft hotel meeting salon.

I really need practical help here, as I am just a freelancer who has been tossed into the waters and told to swim...without any real professional lessons.

The company has hundreds of clients and records thousands of sessions a year. I cant change the way they operate, so at this time I would much rather receive practical advice on specific workflow techniques over a general critique of the underlying situation.

Thank you for any and all assistance you can give.

[lowland]

If you can't change the circumstances under which the recordings are made, then your original post would suggest you're probably making about as much as can be made of a difficult situation. I realise you don't want to hear any of this, but I'm afraid you'll probably just have to dig in and make it happen this time round, and try and exercise more control over what you're given next time.

Specific workflow and editing tips are fine, and others may supply those, but IMO it's the bigger picture that's the more important one. You're hurting right now because you feel you're getting little or no help either from people like myself or your employers, but I assure you that these are exactly the kinds of situations that sort those who are adaptable, inventive and have stick-to-it-iveness from the rest, and that you'll come out of this stronger or find something less stressful to do.

Hang in there, Gary!

[Garynyc]

Thank you for your thoughts and encouragement Nigel!

Anyone else out there who can offer up a streamlined workflow for conference recordings?
 

[ggidluck]

You might want to try Samplitude. Working in the context of a virtual project you can split the regions where each person is speaking into an object. Each object then has it's own properties. You can set the volume level (like normalization) and it doesn't have to be rendered. Any undo is just a setting change (not unrendering of a previous snippet). It is a very good/fast environment in which to work.

I would preview each section on speakers and use a level meter to get the audio into the right ballpark. After a while you will be able to make a quick level adjustment by ear.

Once all your regions are marked for a uniform level if it were me I would run the whole thing through an analog compressor to smooth it all out and get the levels up. If you do this step your initial adjustments don't have to be as precise.

This doesn't have to cost a fortune to make the change. You can get the lesser version of Samplitude and for a compressor something like an ART ProVLA will do.

This is just my approach to the problem. Not necessarily the best or worst, but using equipment and software that I have firsthand knowledge about.

[Garynyc]

Sorry but we are required to use Sound Forge only. Thank you for a possible solution though.

[thechrisl]

After dealing with any clipping, you might want to run everything through a pair of filters (High Pass ~ 80Hz and Low Pass ~ 9kHz) to remove any rumble and edgy treble (+ help with noise).  Then try a limiter plugin with a threshold set close to the lowest talker to get everything roughly at same level.  Normalize after that.  Use some Noise Reduction to help with the noise floor which will be made worse with the limiter.  It won't sound very "natural" but it should get the job done quickly since that's the situation you've been forced into.  I'm pretty sure SF includes plugins that do this but there may be some free ones out there if not. 

Assuming you can get a formula/combination of settings which works pretty well with every scenario, the batch converter would definitely help.  I would think -3 to -1 is fine as a final level.

[SafeandSoundMastering]

My immediate advice once you completed this job is to buy a small mixer with insert points and a couple of low cost compressors to sort this out at source.

Have you just began recording speech?

Being freelance you must invest in some kit man, if jobs go sour you are history.

Seriously you best pick up a little Mackie desk and do it right next time, no joke Behringer kit would have stopped these problems.

You are going to have to use some plug in compression to even stuff out and the clipped stuff is
unlikely to be rectified, just pull it down compress/limit tastefully, High pass it at 60Hz roll off some 14kHz and above.

Don't batch it if you want to work again pop some low DSP use plugins on the output
and render each file with some custom settings, otherwise there might not be a next time, you learnt a lesson here don't repeat it.

I have done absolsutely sh*t loads of location recording in my time and you need the kitchen sink as you are on your own and jobs need to be done right, back ups recorders, UPS, spare cabling, spare everything basically.

Good luck.

[Garynyc]

Well that would be nearly impossible since we have up to 15 concurrent sessions going off at a time. We are given Marantz PMD CF recorders to use, and they arent too bad.

If the CF card/recording fails, it fails--the company does not run backups.

I will take your suggestions seriously.

Thanks Safensound!

[Garynyc]

Safensound can you recommend any low cost compressors?

[Garynyc]

Chris Thank You for these ideas...I will experiment.

[Ed Littman]

Really Nice Compressor 1773
http://www.mercenary.com/realniccom.html

[Garynyc]

Thank you Ed. Unfortunately we get so many rooms sometimes that its almost impossible to get compressors for all of them. Most meetings we do arre well set and with a board op. Sometimes with a cheap client we get the set & forget rooms, different mics, lousy boards etc..

[Treelady]

What you're doing is "essentially" broadcast work, meaning: multiple sources, different processing requirements by source, and killer deadlines.   

There is a reason SADiE and Samplitude are used in broadcast: they were designed to do the things you need to do fast.  SoundForge was an awesome two track editor, but what you're required to do, well its tough to work as fast as you would if you were allowed to use a DAW that was designed with Broadcast requirements in mind.

Sorry to be a downer.

[Garynyc]

Im willing to do most anything to make my daily bread, Tree, even if it means hitting the road with the lowly Sound Forge..

Thanks for your reply!

G

[SafeandSoundMastering]

The RNC will only do stereo, not discrete mono channels, so it is not ideal for working with speech needing mono channels really. In saying that with this particular case you could slap it on the master going to the flash recorders.

We need to keep perspective, low budget and getting a workman like job done, if you had these on 4 channel inserts:

http://www.behringer.com/EN/Products/MDX4600.aspx

Your audio would sound better. This is not critical recording, audiophile work, but basic dynamic range control is sensible and nigh on a pre-requisite. (even on a budget IMO, thats why Behringer exists)

My experience with Behringer equipment is that is it cheap, reasonably reliable over 3-4 years, quite coloured (tad harsh, little roll off of top end) and either works when you get it or it doesn't. FWIW, I actually like the compression action of Behringer Composers, predictable, very controllable, and of a specific character, and I don't really give a rats what anyone else thinks. If I had a Composer that did not muffle audio (as in passed through nice and clean) I would be happy to use one on pretty much anything compared with other cheap compressors like dbx cheapo stuff, Focusrite compounders, TLA, and Pro VLA II's etc. As I understand it they were a copy of a Drawmer compressor design.

I hasten to add the only thing I use Behringer Composers for now is as an active splitter on rare occasion.(and that was the first ones they made which were made very well 15 y.o. and still working )

As the CF machines only have 2 inputs you are sub-mixing mics yes ?

Seems you are tied by the kit you have been given, I respect you for asking how to improve this situation.

cheers

[Garynyc]

Thank you SafenSound;

I dont supply or spec any of the gear we use. We have to deal with what we are given. The only thing we can do to is edit the recorded file in sound forge. I appreciate the gear ideas, but thats not an area as a freelancer that I have any say-so.

 

[SafeandSoundMastering]

Any time G, then your friend is the input gain: set low.

You can de-noise better than you can de-distort.

Your situation is different but when I freelanced in the vast majority of cases I took my own kit, I knew it was maintained to the highest standard, I knew each lead I made was tested, and I knew
the kit was tested and worked before it hit the road.

cheers

[Garynyc]

Roger that, BG!!!

[lowland]

What you're doing is "essentially" broadcast work, meaning: multiple sources, different processing requirements by source, and killer deadlines.   

There is a reason SADiE and Samplitude are used in broadcast: they were designed to do the things you need to do fast.

Yay to that. Shame Gary is stuck with Soundforge - SADiE can be very quick for this kind of stuff, I still do a percentage of speech-based edit work and find it invaluable. I just bought a new mouse (Genius) and keyboard (Microsoft) and have realised that I wasn't aware how much the previous ageing set was slowing me down, things are even slicker now.

[jamesp]

The only thing we can do to is edit the recorded file in sound forge. I appreciate the gear ideas, but thats not an area as a freelancer that I have any say-so.

As a freelancer you should be able to buy your own tools that will do the job. Sound Forge might well do the job in the end but it will take 10 times longer than a more appropriate piece of software. Since you have a copy of Sound Forge, you may also have access to a version of Vegas (it came free with my copy of Sound Forge) which may be more appropriate for what you want to do as you can chain effects and split the file up and put sections on different tracks with different processing.

James.

[Garynyc]

Hi James;

Will Sadie or another application really improve workflow over Soundforge? Can you give me some examples that I may use if I confront our betters with new information?

Thanks!

[ggidluck]

Also Gary you mentioned something about a "tin can" sound. You never went into any detail about microphones. Microphones and their placement are without question the single most important factor in the sound quality.

Let us know what is being used. While this is beyond the scope of mastering if you can make improvements to the audio capture it will make your job easier.

[Garynyc]

Gidluck as I have said in previous posts I'm working all over the country at hotels and convention centers. We don't supply any of the sound gear. We just get a feed from the mixers/patch bays.

[ggidluck]

I was hoping that you could convince this company to allocate some funds to either improve the audio capture or add software that would help streamline your workflow. (In the long run to make your job easier).

[Garynyc]

Gluck they are too busy building up their web presence and internal server infrastructure. Im not faced with an impossible job, Im just looking to get a streamlined workflow going in Soundforge...

What order should I:

• Normalize--What level with RMS?
• Clipped Peak Restoration
• Clipped Peak Detection
• What level should I clip the peaks to?
• Noise Reduction
• Mix from stereo to Mono--Its voice recording
• Any other practical steps Im missing for recording speech?

Thanks for any help people!
[/list]

[Jason Poff]

I've had good luck in similar situations with the waverider plugin found here: 

http://www.automaticmixing.com/


You dupe the track, insert plugin on the duplicate, then assign the plugin to control the fader on the original track.

Then you nudge the duplicate 400ms forward so the plugin has time to respond to the incoming RMS or peak level.  (I believe peak only needs 40ms or 40 samples I can't remember)

Takes some tweaking, but I can generally get something that's set and forget in about 10 minutes.

[Garynyc]

Oh its Mac only. unfortunately we only have laptop PCs to edit with. Thank you all the same...

[Jason Poff]

I believe waves makes a similar product.

[jamesp]

What order should I:

• Normalize--What level with RMS?
• Clipped Peak Restoration
• Clipped Peak Detection
• What level should I clip the peaks to?
• Noise Reduction
• Mix from stereo to Mono--Its voice recording
• Any other practical steps Im missing for recording speech?

Thanks for any help people!
[/list]

Personally I would take a look at Reaper as it costs nothing to try and very little to buy. It may not be quite as efficient as Sadie but I would much prefer it to Sound Forge.

For something like a discussion panel I would open up your stereo file and remove any unwanted sections. I would then split it into sections depending upon who is speaking. I would then move the sections onto different tracks - one track for each speaker. The thinking behind this is that different people will speak at different volumes and possibly need different processing on their voices. You may want to add eq and noise reduction on problematic tracks at this stage too.

I would then use the mixer faders to bring down the louder voices to match the level of the quieter voices. Once all the voices had been roughly matched I would use a compressor on the master output with a long look ahead, a slow-ish attack time to match the look ahead time and a slow release time in order to tame any really loud sections. After the compressor I would add a limiter to catch any other peaks that might slip through and bring the level up to the final level desired.

The final stage is to render everything out to a single file in the format that you want.

James.