HDV's compressed audio/

[Terwingen Niels]

hi,

before i start buying and investing in audiogear for documentary use i should first clear up this next question: is compressed HDV audio (MPEG-1 Layer II ) any good?
i own a canon xh A1 and wanna keep the option of blowing upto big screen.
i dont think is wise to invest in good mics and mixers if you end up anyway with compressed sound? the cam would be the weakest link in the audiochain...

what is your opinion? is hdv sound satisfactionary enough?
or should i get a recorder an deal with the sync problems..(i will be filming lots of footage...)

in case of recorder:
i am thinking of the edirol R44 (i have a limited budget..): should i buy a mixer and go then line into the recorder or should the recorder on his own, without the mixer, be sufficiant?

thanks...

[Gary Nattrass]

HDV sound is fine and 48k 16 bit, I have a Z7 camera and use the on board mic with a radio mic.

Or if I need more I use an ENG-44 mixer hooked up to a sony d-50 flash recorder, I also link the output of the mixer to ch 2 on the camera with a sony uhf diversity radio mic.

The Eng-44 mixer has four inputs so I can have boom and up to three radio mics recorded to the d-50 and also to ch2 of the camera.

Ch1 of the camera also has its own mic recorded so it pretty much covers every eventuality.

I also have two sennheiser g2 radio mics.

[Terwingen Niels]

Hi,

as i understood when you record hd on a tape (HDV), wich is more data then DV, there is less space on the tape to record sound. thats why there is a compression of the audio, much less then the 16 bit 48 khz. in the canon's case they use a compression MPEG 1 layer 2 at 384 kbit/s. wich seems a big compression to me.
i dont know about your Z1 but i guess he is using the same compressionsystem then all the hdv systems. meaning not the 16 bit 48 kHz as you mention...

HDV= 384 kbit/s, compared to 1536 kbit/s for DV video and 1411 kbit/s for audio CDs

my question: how important is this qualitydiffernce (thinking i wanna keep open the option of blowing upto bigscreen..) ?and how are you guys handeling this problem?

[Matt Davis]

Originally Posted by Terwingen Niels

as i understood when you record hd on a tape (HDV), wich is more data then DV, there is less space on the tape to record sound. thats why there is a compression of the audio

In my experience with HDV audio,

- DV and broadcast sound wobbles around -18 dB. With HDV, I let it wobble around -12 dB because HDV sound is compressed and loses some of its dynamic range. Broadcast sound can be pulled up from -18 dB because of good kit and no compression. HDV tends to use cheaper microphones (MK66 et al) and records to a compressed format where headroom and frequency range are compromised to save space.

- MP3 (OKAY! MP2 format! Thank you, Audio Police! But the same rules apply to MP3? You can explain the differences...) compression works because there are some frequencies we can't hear, and the full dynamic range is rarely used. Therefore, if recording to MP3 format, you're not getting full dynamic range, you're best off getting peaks up to -12dB, and Thou Shalt Employ a Limiter.

- IMHO, HDV audio is fine for voice/dialogue. But it is no good for HiFi, 'Atmos' (ambient sound) or especially music and FX. Yes, you can do it, but you can't do much with it once you've got it...

- IMHO, If you're doing true theatrical stuff and can afford a sound engineer, record separate sync sound by splitting the audio from your Sound Engineer to camera and 24bit 96k audio device. You can thank this community later.

- IMHO, If you're doing corprate sound, engage the limiter, use manual, don't rely on camera-recorded audio for anything other than voice. I'm assuming quality microphones...

- NIMHO, If you're doing event work, invest in and use quality microphones - woe betide anyone who relies on inboard mics.

I shoot lots of corporate events, often cut to library music (in the edit, it's 2nd generation from uncompressed). I also get lots of tapes shot on low end cameras...

Sound is worth getting right. PDX-10 will intercut with EX-1 or better if the sound is good.

Best,
M.

[Gary Nattrass]

HDV audio spec here: http://www.hdv-info.org/HDVSpecifications.pdf

Mpeg1 48k 16-bit

[Terwingen Niels]

thanks Matt for the info,
I guess its going to be a NO as I am interested in recording other sounds then only dialog....

quote:
- IMHO, If you're doing true theatrical stuff and can afford a sound engineer, record separate sync sound by splitting the audio from your Sound Engineer to camera and 24bit 96k audio device. You can thank this community later

guess thats going to be a problem with no timecode on my canon xh A1! so what would be a solution in my case? as I am planning to shoot loads of footage in southeast asia..an ediroll 44 and a load of moleskine to make notes every 2 minutes?? the good news is that I will most likely have a sound boom/operator. but she lacks some experience..


I am sorry Gary but I dont understand what you are trying to say.I still read in the chart that its 384 kbits/s after the compression?


greetz

[Matt Davis]

Originally Posted by Terwingen Niels

guess thats going to be a problem with no timecode on my canon xh A1! so what would be a solution in my case? as I am planning to shoot loads of footage in southeast asia..an ediroll 44 and a load of moleskine to make notes every 2 minutes?? the good news is that I will most likely have a sound boom/operator. but she lacks some experience..

You got a sound/boom person? With good mics and a mixer? So long as she rides the levels a little hotter (louder), your sync sound will be fine. I could start a religious war about running audio levels louder, but I hear more of a high 'noise floor' on HDV, so if you have to 'up' it (as one tends to for DVD and have to for web), unless you record HDV at around -12 (and rely on the limiters to prevent distortion if anything untowards happens), your 'upped' -18 to -24 audio seems to have a bit of audio stubble around it.

But if you both record sync sound and room tone at 'around' -12 to HDV, and then record wild track (or any major music numbers) separately to an H4 or something, with a mic tap at start or end and an on-camera audio slate, you really are covered for broadcast.

Only an audiophile will poke an HDV audio recording of voice. But if it were music or anything with a wide dynamic and/or frequency range, an H4 is desirable. <grin>

[Petri Kaipiainen]

In real life it is difficult to distiguish between 384 kbps audio from pure 16/48 wav. Just make a MP3 file at that bitrate and try to tell it apart from CD or similar. Most people happily listen much worse music on their iPods. So this is not the first thing you should worry about your audio, getting a good feed to the system is the first priority. After the mics, placement and levels are perfect and you are still not getting it, only then consider blaming the HDV audio standard.

I suspect some people here get the two compresions mixed up, audio level compression (like in limiter and radio broadcasts) and data compression (like in MP3). The later does not compress audio levels.

Recording 24 bit audio to an external device is of course a wise course of action for many reasons, but using 96 kHz sampling is an extra overkill* and means an extra step in post. At least if you want to hi-fi it for some reason (feel better?) use 88.2 khz which makes the inevitable downconversion more perfect.

Summa summarum: worry about the beginning of the audio chain first, mics, technique, levels, post mixing etc and after perfecting that start to worry about the imagined quality losses in HDV system. Easy to blame the equipment, if they are not theoretically the best possible, but 99.9% of the time the fault is somewhere else.

*) Video mics do not record anything above 20 kHz or so anyhow, so what is the point?.

[Matt Ryan]

if your plans are to make something serious and in some cases "blown up to the big screen" you shouldnt be recording audio into the camera. You should have a professional mixer and boom operator and record the sound externally. Even documentaries and reality tv are shot this way. Sound is the most overlooked aspect in filmmaking and yet one of the most noticeable differences between goos and bad films. Think about it...
If it sounds as good as it looks you are doing a good job. If possible record externally and run an xlr out of the recorder to the camera for syncing and monitoring.

[Peter Moretti]

Originally Posted by Terwingen Niels

...I am sorry Gary but I dont understand what you are trying to say.I still read in the chart that its 384 kbits/s after the compression?...

Yes you're both correct. HDV sound is 16bit/48K and it is compressed. DV sound is 16bit/48K uncompressed.

[Gary Nattrass]

Sorry if I have confused anyone, I have worked in TV and Film for over 28 years and judge things on what it sounds like rather than bit rates etc.

When I started it was all analogue and film was done on mag with at best dolby A noise reduction.

These days the quality of sound on camcorders and such is superb and even i-pod is good quality, forget about compression codecs and just get on with film making, with some care and attention you will get very good results. I also agree that a sep sound recordist will give great benefits and the sort of set-up I described above will hopefully cover all eventualities.

Dont worry about the BIG screen thing I have heard dreadful optical and dolby mis-tracked tracks all over the place and bear in mind that all film prints to my ears have some sort of compression that can be far worse than the front end codecs for HDV etc.

[Petri Kaipiainen]

Amen to that!

I like to add that according to my measurements XH-A1 audio front end is capable of over 90 dB s/n ratio with line in signal and over 80dB with mic in. That is much better than what professional analog Nagra systems can do which were the much envied movei recording gold standard just a decade ago. So no worry about the HDV audio, get everything else in the audio chain working first.

[Steve Oakley]

uhg! Narga. had this fight with a director to record to DAT instead of Narga cause thats what I had. he was all just stuck up about how good a nagra is supposed to sound. after we got back into the studio and sync'd it all up, he goes, wow, can't believe how good it sounds"... oh, I also had to fight with him to use a wireless boom pole too because he was used to low end wireless units instead of lectro's. There was no way I could of used a wired boom on that shoot.

384kbits/sec is 192kbits per channel. MP3's music as 192 stereo is considered good quality. put another way, 384 is about a 4:1 compression which is pretty light. I might add that JVC cameras also record the uncompressed 16/48K as well, but it requires a 2nd pass at the tape to get it.

a good mic, good placement, and good levels will be more important then the recording medium for most purposes. bad room tone, low levels, hum, ect will be more of a problem regardless of whatever is used to record. or put another way, garbage in, garbage out. the better that goes in, the better that comes out.

used analog betaSP audio for the last 15+ years and got by, and its specs where nothing to get excited about, at least the linear tracks. FM tracks where better.

[Pietro Impagliazzo]

How audio compression compares to image compression?

Does audio loose quality just like an image or in a slower rate?

[Steve Oakley]

1. it doesn't

2. yes they both lose quality unless its a lossless codec, but just like video it all depends on how it was compressed - ie what codec as some do much better then others at the same data rate. Apple lossless vs MP3 vs MP2 vs IMA4:1 will each produce different results at a given b it rate. if you can hear the difference all depends on
quality of original recording
quality of the compression - even the same MP3 compression can sound different at the same bit rate due to how good or bad a given encoder does its job
quality of the playback invironment - bad speakers and you won't know the difference, good ones maybe, very good ones will reveal the subtle or not so subtle differences.

so the answer is, it all depends