Print

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

[KingSparta]

Quality Of Sample Rates

 5K (Telephone Audio Quality)
 8K (AM Radio Audio Quality)
11K (Average Mono Voice Quality)
22K (FM Radio Audio Quality)
32K (Digital Audio Quality)
44K (CD Audio Quality)
48K / 96K (Professional / Studio Audio Quality).

Does Anyone know what the sample rate a 33 1\3 And 78 RPM record Would Be?

[KingSparta]

If you recorded a song from FM radio and the sample rate of FM is 22k would there be any need to record this to a media file higher than 22k?

[jgreen]

Answering your last question first, yes.  Dr. Nyquist discovered that, even though a few people hear sound as high as ~22,000 cycles (most of us top out around 15,000), due to harmonics, those people could tell the difference between digital sample rates of 22k and 44k on the same track.  Therefore , 44k is the "Nyquist Number" and anything higher than that (which is TWICE as high as the human ear can hear) is the domain of "audiophiles".

So.  15k is perfectly reasonable, although there may be a certain artifacting as your samples and your cycles overlap.  As you get lower, you're going to lose the highs.  Mind you, we're nowhere close to the range of human voice. 

As for vinvyl, the slower the spin the higher the sample rate, AFAIK. 

[John Gateley]

Hey King, Feeling bored today? Cool questions.

The faster the spin, the faster the sample rate (higher quality): the needle covers more
ground, and so more samples, at the faster spin rate. (Keep in mind I have no idea
what I'm talking about ) Kind of ironic, considering the 78 rpms I had as a kid were
usually poor quality kid songs and the 33 1/3 (or even 16, though I'm not sure I ever
saw one of those) were the high quality music.

j

[KingSparta]

I have been looking for the answers for sometime and can't find an answer.

I would also like to know the quality of any recording equipment back in the 30's 40's and 50's

My point is if it was broadcast over am radio in the 30's-40's the quality was no better than 8000 samples per second and what would be the point in recording from a record or other media from the time period at 44,100 samples.

I know the wow and flutter was no where near .05db it was more like 52.0db

I was thinking if you used 32kbps and 22K Samples this would cover any quality any recording equipment could muster at the time.

I have done some tests of just the look of the wave form and found no real change in the look of the wave form.

I do know when they went from 78RPM to 33 1\3 RPMS it was considered a brake thru But what was the quality?

I also know that some Mic's were horrible and picked up more white noise than they did the singers voice. I talked to a collector last week about Mic's of the 30's and 40's

there was also a 16 rpm record player, this was a fit in the dash for your car record player. it did not make it into too many cars however.

Any Opinions to what type of quality old recording equipment were rated at?

[Noman]

I think you are talking about "sample rate" when you mean "Frequency range" ?!
I think there is no sample rate whith analogue recording...

I have records (bad recordings) from the 70's that gives "flat" frequency response to only 15 khz and I have High quality recordings from the 50's that reach over 22 khz.

Playing back vinyl also gives a lot of "overtones" up to way beyond what you're able to hear.

[Alex B]

Vinyl and FM don't have a sample rate because they are analog.

The frequency range of a high quality cartridge can be 20-25000 Hz or even bigger, like e.g. this manufacturer claims: http://www.ortofon.com/html/profile.asp?body_magnetic_technical_data.asp. FM audio has a more limited frequency range that is from 30 to 15000 Hz for technical and historical reasons.

An analog audio signal "gets" a sample rate when it is sampled for saving its representation in a digital format. To represent a 20 kHz signal in a digital format a 40 kHz sample rate is needed at least. More info about sampling is available for example here: http://en.wikipedia.org/wiki/Nyquist-Shannon_sampling_theorem

When the Red Book CD audio standard was defined the 44.1 KHz sample rate was chosen for a few reasons. One of the reasons was to be able to reproduce a 20 KHz audio signal with some safe margin.

For FM audio a 32 kHz sample rate would be enough.

If the audio file is digitized from an old low quality source you can check the actual frequency range for example with Spectrum Lab audio signal analyzer. It is freeware and actually the best that I have seen.

With LAME encoder you shouldn't worry too much about the best settings. It selects automatically the best parameters and lowers the sample rate when suitable. A setting like --preset 32 -m m should be about as good as possible for an old mono source if 32 kbps is preferred.

Edit: Actually my next post has a better LAME setting, which produces usually about 32 kbps speech files, but can use a higher average bitrate when needed for keeping constant quality.

However, MP3 starts to produce ugly audible artifacts already at much higher bitrates than 32 kbps. These compression artifacts can make overall audio quality unpleasant even the theoretical bandwidth is wide enough (like up to about 10 kHz when a 22.05 kHz sample rate is used).

[KingSparta]

I will look into that program later, thanks for the links\Info

Since most Old Recordings I Work With On A Normal Bases Are Basicly Radio Plays And Not Music For The Most Part, Would That Also Not Have An Impact On How Low You Can Go Without Having "ugly audible artifacts"?

Would 4 People Talking In A Radio Play Need Such A High Samples Per Second Or Kbps Bit Rate?

[BullishDad]

King Sparta said:

there was also a 16 rpm record player, this was a fit in the dash for your car record player. it did not make it into too many cars however.

This made me curious so I googled and found this interesting article about the invention, including a picture of the "Hi-Way Hi-Fi".

http://www.imperialclub.com/Repair/Accessories/HiWay/invent.htm

[jgreen]

Sparta--

I believe you are on the right track, and I would suggest you try wiki P using search terms such as "att+samples", etc.  It was ATT who did much of the trial-and-error tests on audio sample rate, and that is why digital phone streams are encoded at 5k/I-don't-know-what-bitrate.  Also, try googling for xm and sirius.  Both sat radio networks apply varying bitrates to their streams, depending on content.  Talk is very low, while orchestral music is their highest.

Also, if you're going to encode at lowend bitrates, look into AAC+ (Alex B is an expert in this, as opposed to analog sample rates).  AAC+ uses a logarythmic distribution of sample points vs frequency, I believe, giving you better bang for the buck.

In saying all this, I bring to bear all the technical credibility that a degree in English Literature conveys.

[JimH]

In saying all this, I bring to bear all the technical credibility that a degree in English Literature conveys.

You have a Masters in Understatement.

[Alex B]

I will look into that program later, thanks for the links\Info

Since most Old Recordings I Work With On A Normal Bases Are Basicly Radio Plays And Not Music For The Most Part, Would That Also Not Have An Impact On How Low You Can Go Without Having "ugly audible artifacts"?

Would 4 People Talking In A Radio Play Need Such A High Samples Per Second Or Kbps Bit Rate?

Normal human voice reaches only about up to 3-4 kHz, so one would like think that a 8 kHz or 11 kHz sample rate would be enough. However, even speech recordings should contain enough higher harmonics to make the speech sound natural.

I have found that for example LAME encoder doesn't work well with sample rates below 22.05 kHz, which may be an optimal sample rate for low bitrate encodings. In last July you posted a 128 kbps mp3 radio show and asked something about encoder settings for reducing the bitrate to 32 kbps. I tested an optimal setting (in my opinion) and was about to post a converted file, but then I saw that you had already used GoGo at default 32 kbps CBR.

With the default setting GoGo uses 44.1 kHz stereo which is highly unoptimal for a low bitrate MP3 file. GoGo was developed to be a fast alternative for LAME at high bitrates when the reduced quality does not matter much. It is not optimized for producing high quality.

My file has clearly a lot better speech quality than a GoGo encoded file, but I was afraid that you already deleted a bunch of your original files and didn't post my file because I didn't want to make you upset.

I just accidentally found the folder again and instead of deleting it now I decided to post the file and the settings I used. I hope this does not upset you now.  

GoGo 32 kbps, MC custom setting: gogo.exe -b 32
SupermanGoGo32.mp3 (5.97 MB)

LAME VBR, MC custom setting: -V8 --vbr-new --lowpass 9.2 --resample 22 -m m
SupermanLAMEVBR.mp3 (5.96 MB)

[KingSparta]

BullishDad

Yep, there she blows...

It Must not have done well on In Testing On Curves, Hills, Dips, Pot Holes, Rail Road Tracks, Stopping, Take Off's and Straight-A-Ways

[KingSparta]

LAME VBR, MC custom setting: -V8 --vbr-new --lowpass 9.2 --resample 22 -m m

Is this also avail in English, Or Just Yiddish?

What Are The Effects On A 32Kbps File Re-sampled From 16k To 22K?

I would think it will just take up more space on the disk with little or no degradation of the media file.

[Alex B]

Is this also avail in English, Or Just Yiddish?

 

Use the VBR quality level 8, use the new VBR mode, use a lowpass filter at 9200 Hz, resample the file to 22050 Hz (if not already 22050 Hz), use mono mode

However, LAME understands only Yiddish. The actual command line string is literally

-V8 --vbr-new --lowpass 9.2 --resample 22 -m m

lame.exe in front of it is not needed in the custom dialog box. MC uses it by default. GoGo needs gogo.exe.

What Are The Effects On A 32Kbps File Re-sampled From 16k To 22K?

I would think it will just take up more space on the disk with little or no degradation of the media file.

It would be about twice as bad. It would go like this:

- Decode a highly compressed, possibly already quite bad file, to uncompressed 16 kHz PCM
- Resample the file to 22 kHz (which adds some distortion)
- Encode the file to MP3 (thus reduce a lot more information since the encoder starts with the uncompressed 22 kHz PCM data)

If the source bitrate is not at least twice the target you should avoid re-encoding if possible.

[KingSparta]

use a lowpass filter at 9200 Hz

What Is This For?

[Alex B]

A smooth lowpass filter just below the highest possible frequency (=11 kHz at a 22 kHz sample rate) produces usually a bit better quality.

By default LAME -V8 mono uses a 32 kHz sample rate and a lowpass filter at 12.5 kHz. This would typically produce a much higher average bitrate than 32 kbps with a speech file.

That's why I reduced the sample rate and the lowpass filter frequency. The resulting VBR file is 32 kbps and has decent quality.

LAME -V9 (a step lower VBR quality setting) would have produced a bit lower overall audio quality in this case.

[KingSparta]

A smooth lowpass filter just below the highest possible frequency (=11 kHz at a 22 kHz sample rate) produces usually a bit better quality.

And now a question I always asked my Mom when i was a kid

Why?

Why Would I Want To Use A Filter?, What Does It Filter?, And Why Would I Want To Filter It?

The Cable guy put a filter on my cable once, that was not good, I had to take it off to watch Aliens back in the 80's on HBO. If i use this filter will i now missout on Friday the 13th Part 59? If not what will I be missing out on?

[Alex B]

I used it because Mr Lame said to me (in Yiddish): "-V8 won't fit".

(The quality issue with the abrupt high cut at the highest possible audio frequency is difficult to explain shortly. It is a common belief in audio industry that a smoother lowpass filter below that frequency produces better quality.)

[BullishDad]

BullishDad

Yep, there she blows...

It Must not have done well on In Testing On Curves, Hills, Dips, Pot Holes, Rail Road Tracks, Stopping, Take Off's and Straight-A-Ways

Actually, it did well on the torture test:

Kent shot the car over the trestles, but there was not even a waver in the sound. He ran over cobbles, skidded past wash-boards, climbed up and down small, jutting mounds. Still the music came forth, loud and undisturbed. Kent was impressed and immediately said he would demonstrate the set to the president of Chrysler.

There were problems when Chrysler put the product in Plymouths and Dodges.  They were not built as quiet as a Chrysler, but they eventually worked that out.  The real downfall was the lack of software.   Not enough 16 2/3 RPM disks were available and the marketing campaign dried up.

Worth the full read if you like gadgets.