Alright, I have had enough of the ultra wide volume levels broadcast via cable television. They are most notable when commercials are on and sometimes from station to station, but can even be a problem with a single television show or DVD rental.

What I'd like to do is "normalize" the TV audio output before it reaches the speakers.

I know there are programs out there that can take a single song or audio track and normalize the track with specified parameters. There are others that can normalize multiple tracks, such as mp3's, so that they are consistent from track to track. This is what I'd like to do real-time with the TV audio output.

For the most part it can be an external device as I usually pass my TV audio to my amplifier and to external speakers.

The part that I am most troubled by is this. Any any given volume setting, the output volume to the speakers changes. Normalizing to 80- 90% or so is fine, but when I manually want a higher or lower volume, how would the circuit detect this and then output a new 80 - 90% normalization?

I have no proposed circuit for this idea as I an clueless on how to achieve this. It could quite possible involve microprocessors, and there could very well be some audio chip that does some or all of this already.

Thus.....I am looking for any and all suggestions on how to achieve this goal.

Thanks!

 

What you want is an audio AGC. Not hard at all, just google "audio AGC". Bet someone comes up with something easy before you read this though.

 

Thus.....I am looking for any and all suggestions on how to achieve this goal.

I know what you wants, as I want the same thing.

An AGC is not the solution. On program material with people speaking & pause, very soft music or complete silence, the AGC will boost up the gain to an unacceptable HIGH level then nearly mute it to an acceptable LOW level before recover.

A human being will not do that. He/she will only turn down the volume when it is too loud.

What you want is to make a device to emulate a human being, which:

1. detect how loud something is and act upon accordingly
2. does not increase the gain if the volume is low or between pause
3. does not increase gain if there is no signal

Only a device under programming control can achieve that.

 

Hello,

A good AGC control is well defined at attack and decay delays.
It will act on rush load noise and let it come back slowly.

Take a look at this PDF : http://www.eecg.utoronto.ca/~kphang/papers/2001/martin_AGC.pdf

It comes from this page of the EDUCYPEDIA.
http://www.educypedia.be/electronics/amplifiers.htm

Greetings,
Bertus

 

I managed to cook up a circuit that can rectify the input signal and give me basically the envelop of the peaks. I tested it and its output looks OK. I can read the instantaneous value of this envelop voltage into the microcontroller using an ADC pin. With this I can do some limit checks on the ADC results or build a rolling average.

I can also digitally control the gain of an amplifier downstream via the uC using digital pots.

My goal is to emulate a real human being w.r.t. adjusting volume control on the signal.

Problem is I have little idea on what to do with the sampled data. How can I work out when to increase or decrease the gain or just stay put?

 

You can use an analog audio limiter. Follow that with an amplifier, if you choose, to adjust the output level for various ambient conditions.

 

eblc1388,

Thanks for your input. Normally Your last post might be considered as hijacking
a thread, but since your goal seems so closely related to what I am looking for I am hoping your progress is also my progress. Maybe you can share more of your design when something is complete.

I have downloaded a couple of datasheets on actual IC's that are capable of
accomplish this. I haven't read them yet but they look promising.

 

An NE570, NE571 and NE572 are audio expander/compressor/limiter ICs.
Philips has an Audio Compander Cookbook on their site with all the details.

I have a Toshiba TV with an audio compressor setting. It makes all loud sounds the same volume and the first sounds of each syllable sound like they hit your ears with a hammer. Yes, it makes loud commercials sound louder.

An expander does the opposite of the compression used for loud commercials. But its setting must match the threshold of the compressor which never happens. Then when somebody yells it is really loud.

 

eblc1388,
Thanks for your input. Normally Your last post might be considered as hijacking thread, but since your goal seems so closely related to what I am looking for I am hoping your progress is also my progress. Maybe you can share more of your design when something is complete.

I'm sorry. You're quite right, it is hijacking. I was going to ask exactly the same question in a post if you have not posted your question earlier. Instead of asking the same thing twice and have two threads going, why not just place them in a single thread. Maybe in this way we can get more responses.

What we are facing is not an easy problem to solve. Any musical material if brought to the same audio level will sound extremely dull and no fun to listen to anymore. This is apparent also from the remarks made by of audioguru.

It is difficult because the hardware is there to let us control the overall gain easily but what we need is an algorithm on when to change the overall gain. This does not has to be frequent like a normal AGC but it need to be sensible and effective. When I change the volume on the TV, I just do it once in a while and the result is fine. I just don't want to do that too often because each time a stupid bar appear on the TV to show me what the volume is.

 

The IC I was looking at was by intersil, some applications are located here:


http://www.intersil.com/cgi-bin2/MsmFind.exe?CFGNAME=MssFind.cfg&AND_ON=Y&ALLCATS=X&EN=X&NO_DL=X&QUERY=AN1016


Not sure if it does what we want or not.