7
$\begingroup$

I would like to know if in television broadcast high end audio frequencies are transmitted e.g. frequencies higher than 14 kHz. What is the cut off audio frequency of transmission in television broadcast?

I am preparing a report (for research) characterizing audio frequency response of various devices e.g. iPhones, Samsung Galaxies, various televisions. While doing this, I got curious about the spectrum output of television. Given a piece of Audio for broadcast transmission, how it's spectrum get altered? At what frequency there is cutoff and specially can the cutoff be below ~20K, the human audible range?

Broadly my interest is in knowing what happens to an audio given to a broadcasting channel and by the time it arrives as an input to a television.

Thanks.

Improve this question
$\endgroup$
5
  • $\begingroup$ Analog (NTSC or PAL) or digital (HDTV) broadcast? $\endgroup$
    – hotpaw2
    Commented Nov 18, 2016 at 7:11
  • 1
    $\begingroup$ @user3005720 it's crucial to add this info to your question! Also, maybe you want to give us a bit of background on why you're asking. It's not that easy to give you the right information, because we're looking at a whole system – especially for analog systems, what was sent is not necessarily what is seen by the receiver, and that is definitely not what will be reproduced! $\endgroup$
    – Marcus Müller
    Commented Nov 18, 2016 at 11:04
  • $\begingroup$ as said, edit your question and add that info. We can then address it. $\endgroup$
    – Marcus Müller
    Commented Nov 18, 2016 at 19:56
  • $\begingroup$ and, any system will have a cutoff. There can physically be no system with infinite bandwidth! $\endgroup$
    – Marcus Müller
    Commented Nov 18, 2016 at 19:57
  • $\begingroup$ Shouldn't this question be tagged with a [television] tag? $\endgroup$
    – Stevoisiak
    Commented Jul 21, 2017 at 14:11

2 Answers 2

Reset to default
7
$\begingroup$

In digital television, I think they use mostly a 48 kHz sampling frequency, so it is up to whether the audio encoder allocates any bits to the ultrasonic frequencies:

Before encoding
Figure 1. Before encoding: Left) almost full scale white noise, Center) almost full scale pink noise, Right) full scale 20-24 kHz band pass white noise.

After encoding
Figure 2. After encoding in MPEG 1.0 layer II, 224 kbit/s, 48000 Hz stereo (known to be used in DVB-T) and decoding, using mp2enc for encoding and mpg123 for decoding.

With white noise, in the 16-20 kHz range only some bursts of frequency survive, but above that nothing. Starting with "pink" noise, everything above 15 kHz gets stripped away, because more bits are allocated to the higher-intensity lower frequencies. Starting with content only >20 kHz, some ultrasonic frequencies survive, as no bits need to be allocated to lower frequencies.

Also other encoding methods than MPEG 1.0 layer II are in use in broadcast television.

Improve this answer
$\endgroup$
4
  • $\begingroup$ Nice answer! On the other hand, while valid now, it might change really soon – the ultrasound tagging/tracking might give content providers enough incentive to use a codec that can transmit much higher frequencies for inaudible watermarks. $\endgroup$
    – dtldarek
    Commented Nov 18, 2016 at 12:16
  • $\begingroup$ @dtldarek: Anti-piracy watermarks are not outside the audible range, for the simple reason they'd simply be stripped off. For more decent uses, the problem exists across the whole chain: distribution, speakers, microphones. All perform badly at inaudible frequencies. $\endgroup$
    – MSalters
    Commented Nov 18, 2016 at 15:05
  • 2
    $\begingroup$ @MSalters I was talking more about things like this and this. $\endgroup$
    – dtldarek
    Commented Nov 18, 2016 at 17:13
  • $\begingroup$ @dtldarek Time to wear headphones for everything? $\endgroup$
    – JAB
    Commented Nov 18, 2016 at 18:40
5
$\begingroup$

Digital TV

Aside from the usual thing that no system designer will use a sampling rate that's unnecessarily high, because real humans simply can't hear over ~20kHz, and thus, building DACs and amplifiers for frequencies higher than that won't happen, DVB-T specifies this in ETSI TS 101 154, chapter 6, "Audio".

Olli has already given you an excellent, fact-and-spectrum based review on what gets really through on such codecs, so here's but a bit of standardization facts:

MPEG1/2 audio:

The audio sampling rate of primary sound services shall be 32 kHz, 44,1 kHz or 48 kHz. Sampling rates of 16 kHz, 22,05 kHz, 24 kHz, 32 kHz, 44,1 kHz or 48 kHz may be used for secondary sound services.

(note the decimal "," mark – that tells you a lot about the countries that were not part of the standardization)

As you probably know, for real-valued signals, the sampling rate must be more than twice the highest signal frequency to allow for perfect reconstruction – in other words, the maximum sampling rate of 48 kHz will yield a maximum audio frequency of 24 kHz.

AC-3, DTS Audio

Not really specified in the standard, but sampling rates > 96kHz are yet to be seen in the wild for these formats, generally.

In other words, maximum audio frequency according to these standards would be something like 48kHz, but I doubt that. You'll see the same 20kHz cut-off.

MPEG-4 AAC in all its relevant variants

There's only the MPEG4 profiles specified. But these are generally only compared to 44.1kS/s CD audio – so I guess, I'd expect a TV set to handle 44.1kS/s, 48kS/s, and everything else would be a bit of a hopeful thing, so broadcasters won't use it.

Analog TV

You're joking, right? That's basically the same audio as FM broadcast radio.

NTSC and PAL systems (technically, PAL and NTSC are just Video/color specs, but they are always used in systems that also broadcast audio, so it's kinda OK to confuse the two) use what used to be called "F3E". That's simply frequency modulated audio.

For PAL systems, the overall FM frequency deviation is 50kHz, and for NTSC systems, it's 25 kHz, both with a preemphasis time-constant $\mu_\text{PAL}=50\text{ µs}$, $\mu_\text{NTSC}=75\text{ µs}$. Now, from this frequency deviation we can't really read the contained signal's bandwidth all that easily (we can whip out our Bessel table, and try to find the 1% bandwidth, but honestly, that won't yield any relevant results. Analog FM audio was never transmitted for maximum bandwidth, but for "let's compensate the modest mess FM receivers are", so any rule of thumb will be just as good). Rule of thumb says you don't get any more than 20kHz of audio bandwidth, and preemphasis (and de-emphasis) won't be nicely optimized to retain anything over let's say 75% of that, so you're happy if you ever get 16kHz of usable audio.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ "(note the decimal "," mark – that tells you a lot about the countries that were not part of the standardization)" I think it tells less than you think. They simply made the decision to use that notation instead of the US notation, it doesn't say a thing about the countries involved except some of them are likely European. $\endgroup$
    – Mast
    Commented Nov 18, 2016 at 13:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.