Audio Sample Rate

[mercer]

so, i have been messing around with audio recording options. thus far, i have been using an olympus field recorder at 24bit/96khz and have been getting good sound with a lav.

A friend of mine recently tested audio with an olympus voice recorder, recording at high bitrate MP3 with a neewer lav and the quality is amazing.

I wouldn't say it's any better than my wav recordings but it sounds just as good.

So, my question is... are 24/96 recordings overkill?

A lot of inexpensive voice recorders record 24/48 or 16/44.1, is this level good enough? I even noticed that mpeg streamclip's highest, audio transcoding setting is 48khz, should i be taking the hint?

my higher end olympus field recorder's still fetch a good price on the used market, if a lower bit and sample rate is good enough, i could have 5 lav setups instead of two. 

Thoughts?

[jax_rox]

I'd go with 24/48 wav at minimum.

None of the post audio guys I've worked with would be too happy if you delivered mp3s as the location audio masters.

[mercer]

Yeah, I would definitely do wav recordings but is 96 overkill? I read that depending upon the audience's listening equipment, anything over 48 could actually make your audio sound worse... Of course this article was geared towards music, which I assume would be more demanding due to the higher frequencies. But there is something interesting about having a half dozen small voice recorders dispersed across a set, at actor's marks, plus the ones in the actor's pockets. But I'll just probably stick to what I have... It's working so far. 

[gh2sound]

24bit 48k please

[mercer]

24bit 48k please

So, you think that is more than enough for low budget, indie movie dialogue?

[jcs]

For spoken dialog, 16bit 44.1kHz MP3/AAC 128kbps or higher is plenty good. Here's how you can prove it to yourself: record a test at 96kHz 24bits WAV. Convert this recording to 128kbps 44.1 MP3 or AAC. Do all post work on the recordings, then export for delivery- H264 will use AAC. Can you hear any difference?

Everything we hear now is AAC or MP3, except for the movie theater and BluRay. Far more important than WAV vs. AAC/MP3 is the quality of the preamps (Sound Devices is my favorite- best bang for buck), mic, mic placement, recording level, and ambient noise level. For music and Foley fx, higher rates and uncompressed can be helpful, though the final mix will be compressed in most cases for delivery.

If everyone listened using Stax headphones and B & W speakers, things might be different ;)

[mercer]

For spoken dialog, 16bit 44.1kHz MP3/AAC 128kbps or higher is plenty good. Here's how you can prove it to yourself: record a test at 96kHz 24bits WAV. Convert this recording to 128kbps 44.1 MP3 or AAC. Do all post work on the recordings, then export for delivery- H264 will use AAC. Can you hear any difference?

Everything we hear now is AAC or MP3, except for the movie theater and BluRay. Far more important than WAV vs. AAC/MP3 is the quality of the preamps (Sound Devices is my favorite- best bang for buck), mic, mic placement, recording level, and ambient noise level. For music and Foley fx, higher rates and uncompressed can be helpful, though the final mix will be compressed in most cases for delivery.

If everyone listened using Stax headphones and B & W speakers, things might be different ;)

Ok thanks, I am a little ignorant when it comes to audio... What does the kHz and the bit number represent... Or measure?

[mercer]

The next question then becomes... Are you gaining anything by recording in wav, when it will be compressed later for delivery anyway? Why not just record at a high bitrate MP3?

[trafficarte]

Starting with (too much) compressed audio is like shooting with (too much) compressed video instead of using better codec. If you have to tweak something in the edit room you'll risk to have a lot of issue.

Audio is good enough for most video works at 24bit/48KHz, I've sometimes used 96KHz for music, but it's overkill if you audio equipment quality is less than stellar.

Audio files are a lot smaller than video, so 24bit/48KHz is not such an issue, IMHO. 16bit is problematic if you have to do almost nay tweaking in post: noise reduction, heavy EQ and so on...

[gh2sound]

Don't record in MP3 format for any broadcastable material - MP3 quality audio present in a mix can fail QC (in TV land) - 24 bit 48k is used in 99.99% of every film (AAA titles) and TV mix I've done - music, sometimes at 96k if you're feeling extravagant!

[mercer]

Don't record in MP3 format for any broadcastable material - MP3 quality audio present in a mix can fail QC (in TV land) - 24 bit 48k is used in 99.99% of every film (AAA titles) and TV mix I've done - music, sometimes at 96k if you're feeling extravagant!

Okay, what does the 24 bit number pertain to? And also the 48khz number. What about a lower wav recording... 16bit 44.1 or even lower wav recording? Is the bit number similar to the dynamic range number in video?

[mercer]

Sorry for all the questions, I get that 24 bit 48khz is ideal. But I have an upcoming shoot that takes place at a dinner party. The camera is going to float through the room with my two main characters, but I want to hear snippets of background conversation. It's important to the scene. My plan is to have as many as a dozen audio recorders in the actor's pockets. There are some inexpensive voice recorders that record wav files with the bitrate numbers slightly higher than what an MP3 would record at it's highest settings. 

[Zak Forsman]

Okay, what does the 24 bit number pertain to? And also the 48khz number. What about a lower wav recording... 16bit 44.1 or even lower wav recording? Is the bit number similar to the dynamic range number in video?

​to put it in camera terms, bit rate is like dynamic range and sample rate (kHz) is like frame rate. 24bit has a range of 144db while 16bit has a range of 96db. the latter is more limiting in terms of how from you can swing between your lowest lows and highest highs. But just like how cameras also have a "useable" dynamic range, the noise floor in an audio recording device is also a limiting factor. either way, 24bit gives you more room to work with.

a sample rate of 48khz is adequate for dialogue, but if you were to be recording sound effects that you might want to slow down in post and manipulate to create new sounds, you'd want to record them at a higher sample rate (higher resolution) to capture more detail in the audio. in camera terms, it's like if you wanted a slow motion shot. if you shot at 24fps and slowed it down to half speed in post, you'd be doubling frames. but if you captured it at 48fps and slowed it by half, it would still be smooth in a 24fps sequence.

[mercer]

​to put it in camera terms, bit rate is like dynamic range and sample rate (kHz) is like frame rate. 24bit has a range of 144db while 16bit has a range of 96db. the latter is more limiting in terms of how from you can swing between your lowest lows and highest highs. But just like how cameras also have a "useable" dynamic range, the noise floor in an audio recording device is also a limiting factor. either way, 24bit gives you more room to work with.

a sample rate of 48khz is adequate for dialogue, but if you were to be recording sound effects that you might want to slow down in post and manipulate to create new sounds, you'd want to record them at a higher sample rate (higher resolution) to capture more detail in the audio. in camera terms, it's like if you wanted a slow motion shot. if you shot at 24fps and slowed it down to half speed in post, you'd be doubling frames. but if you captured it at 48fps and slowed it by half, it would still be smooth in a 24fps sequence.

Okay, that makes sense. So, is there any truth to what I read that 96 kHz can actually cause distorted sounds? If recording 96 kHz, do I still want to keep my dialogue peaking at -12, or does that change?

Next question, I am using an older model Olympus field recorder. I just finished editing a short and I was pleased with the sound quality. I recorded it at 96/24. After reading that sometimes recording too high can be detrimental, I looked at my recorder and found my options to be

1. 96/24 

2. 88.2/24

3. 48/16

4. 44.1/16

5. 44.1/mono

Since I do not have a 48/24 option, should I just stick with the 96/24?

Also, I wrote in a previous comment that I have an upcoming shoot that requires a lot of movement and a lot of audio. I was going to set up my main two actors with the Olympus recorders I have, but I also was thinking of getting inexpensive voice recorders for the background players. To keep the cost down, I have found an inexpensive recorder that records wav files but the specs are only slightly higher than the highest MP3 recording settings. I think it's 24/396. Will this audio be usable because it is a wav file or is it basically like using an uncompressed equivalent to MP3 quality?

Again sorry about all the questions and I really appreciate any info. 

[jax_rox]

Stick with 96/24. And keep your levels the same (-12 peaks). I've never heard of extr adistortion when recording 96KHz.

Post up the specific model of recorder and it will be easier to tell whether or not it will be fine to use.

 record a test at 96kHz 24bits WAV. Convert this recording to 128kbps 44.1 MP3 or AAC. Do all post work on the recordings, then export for delivery- H264 will use AAC. Can you hear any difference?

​Converting 24bit 96KHz audio to 128kbps mp3 will actually sound better than starting with a 128kbps mp3 recording. Advising to record in mp3 is silly and you definitely will run into problems if you attempt to do any kind of audio post on it. If you had no other option, then sure - record to mp3, but when one has the option of recording high bit and sample rate wav files, why would you suggest doing anything otherwise?

As has already been said, mp3 quality audio will likely cause QC fail.

[mercer]

This is the one I was looking at for background actors. In the shot I will be following and/or they will be pushing the cam, on a steady cam, through a party. The conversation the main two actors wil be having will be commentated by the other background actor's, but in a comical way... Meaning they may say something about liking a guy and as they pass a group of people, a background actor, having a totally different conversation will be heard saying, "not a chance." I don't have access to a boom operator skilled enough to follow the flow and move the mic back and forth, so I want to do the old recorder in pocket and lav trick. I could need anywhere between 6 to 10 actors for these background characters... That's a lot of money for quality recorders, so I was hoping something like this might work and not sound too horrible...

http://www.bhphotovideo.com/c/product/1004584-REG/philips_dvt_1100_digital_voice_tracer_recorder.html

 

[gh2sound]

Slightly off topic but fun - have a go at this:

 

http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality

[sunyata]

Ever since the whole Neil Young thing I've been wondering about this too and was first inclined to agree Neil, who I'll always agree with in fender amp selections, but in this respect it seems that he might be wrong, 16bit 44.1kHz  is about all the human ear can hear, 24bit 48khz is more necessary for recording and mastering overhead, but not for final delivery. Format wav vs aiff etc is not as important as the codec, i.e. pcm_s16le etc. In a terminal (if you have ffmpeg installed) you can see the audio codec options by typing: "ffmpeg -codecs | grep DEA" - this will give you a list of supported encoding and decoding audio codecs. 

Best reference I've found on this topic, which I think was also in response to "pono". http://xiph.org/~xiphmont/demo/neil-young.html 

Specifically with respect to the bit rate issue from the article link above:

When does 24 bit matter?

Professionals use 24 bit samples in recording and production [14] for headroom, noise floor, and convenience reasons.

16 bits is enough to span the real hearing range with room to spare. It does not span the entire possible signal range of audio equipment. The primary reason to use 24 bits when recording is to prevent mistakes; rather than being careful to center 16 bit recording-- risking clipping if you guess too high and adding noise if you guess too low-- 24 bits allows an operator to set an approximate level and not worry too much about it. Missing the optimal gain setting by a few bits has no consequences, and effects that dynamically compress the recorded range have a deep floor to work with.

An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there's no reason to keep more than 16 bits.

 

[jcs]

Stick with 96/24. And keep your levels the same (-12 peaks). I've never heard of extr adistortion when recording 96KHz.

Post up the specific model of recorder and it will be easier to tell whether or not it will be fine to use.

​Converting 24bit 96KHz audio to 128kbps mp3 will actually sound better than starting with a 128kbps mp3 recording. Advising to record in mp3 is silly and you definitely will run into problems if you attempt to do any kind of audio post on it. If you had no other option, then sure - record to mp3, but when one has the option of recording high bit and sample rate wav files, why would you suggest doing anything otherwise?

As has already been said, mp3 quality audio will likely cause QC fail.

I write code to process audio in real-time, including converting back and forth between the time and frequency domains for effects, pitch shifting, etc. We did tests at Myspace Music to see how many times we could recompress sung recordings before there was a noticeable degradation in audio quality, starting at 128kbps 44.1kHz. I think we got to 9 rounds before we stopped testing (still good enough at that point). After the first round of psycho-acoustic frequency dropping and noise shaping (as occurs with MP3 and AAC), there's not a lot of change on subsequent round trips. This is especially true for spoken human voice which has very limited bandwidth needs.

Will uncompressed 24-bit 96kHz (even 192kHz) sound better than compressed? Absolutely! But only if using really nice amps, speakers/headphones, and when the audio has never been compressed. My Pioneer  RX-A3010 receiver has a feature which 'reconstructs' compressed audio, and it does produce a nicer result when enabled (everything except BluRay (uncompressed 24-bit, etc.) benefits from this feature).

When content is ultimately delivered as H.264, the audio will be psycho-acoustically compressed with AAC 44.1- or 48kHz. For spoken dialog you'll have a hard time hearing any difference vs. original uncompressed recording. We've done the tests, you can too- no need to trust anyone's opinion: easily testable for yourself. If you have the option to use a Sound Devices or anything else in the price range or cheaper, the SD is your best choice (amazing preamps and limiters). However if your budget is less and all you have access to is MP3/AAC recorders, they'll work fine for dialog.

[jcs]

Slightly off topic but fun - have a go at this:

 

http://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality

​Very on topic- not easy to hear the difference! A great test (yes, with headphones and critical listening, but the average person won't be able to tell- that's why almost all audio is now delivered compressed). As noted recording uncompressed 24-bit and 48kHz or higher is the standard as it helps options in post, for dialog that will ultimately be delivered compressed, MP3 and AAC are fine if the budget doesn't allow for uncompressed. Far more important is mic, mic placement, preamps (low noise), and possibly limiters (for unpredictable levels or occasional loud bursts (allowing the overall recording to be closer to -6dB (1/2 max voltage level: -12dB is only 1/4 max voltage: signal to noise ratio is lower). When final audio is normalized, it's pretty close to 0dB (peaks hit max scale, but don't clip)).