MP3 Players: How Digital Data Becomes an Analog Signal

[aeterminator1]

my doubt is , how does a digital data is converted into frequency varying analog signal.
that is from an adc we get amplitude of the analog signal,but what causes the variations in frequency?

 

[mgb_phys]

You don't need to measure or generate frequency.
If you measure the sound level at regular intervals you capture all frequencies (upto half the sampling rate) then when you play it back you are simply setting the position of the speaker cone to a certain level at each time interval.

To picture this, just draw some waveform on graph paper and image in that you only measure it at each square. then draw a line between those squares - this is what the player plays back.

 

[aeterminator1]

Thanks for the reply, what does sound level refer to? Is it the amplitude.
Also,the sampling rate is taken as twice of the highest frequency (since signal contains complex signals),the will the reconstructed signal have this frequency?

 

[mgb_phys]

Yes, ultimately all you measure is the voltage from the microphone and all you output is a voltage to the speaker.
The highest frequency in the (arbitrary) signal you can record is half the sampling frequency. that's the reason that CD uses 44kHz, human hearing goes upto about 20kHz (if you are young enough).

 

[aeterminator1]

Than is the output of constant frequency, irrespective of the input signal frequency?

 

[mgb_phys]

Frequency isn't really a useful quantity here.
You have a data stream (the music)
You can describe this a time varying sequence of voltage levels from the microphone - this is essentially what an audio CD does.
Or you can take the whole data, Fourier transform it, and describe it as a sum of pure sine waves of a different fequencies and amplitudes (this is partly what MP3 encoding does)

There is an input/output frequency in terms of the sample rate but this isn't the same thing as the frequencies in the signal - although the rate does limit what frequencies you can record.

 

[minger]

An interesting thing to look up is the Nyquist Limit. This is what mgb was referring when he mentioned "...up to half the sampling rate..". You'll notice that most digital audio is sampled at 44.1kHz, which turns out to be about twice the highest frequency typically heard by humans (typically we can hear up to about 20kHz).

So as long as we sample at least at 40kHz, we "capture" all frequencies that we can physically hear.

 

[vaibhavgupta]

though i don't know that in what respect u are saying this,
but to convert a digital data into its corresponding freq. u can use digital frequency counter.

 

[MATLABdude]

though i don't know that in what respect u are saying this,
but to convert a digital data into its corresponding freq. u can use digital frequency counter.

Maybe if the original poster enjoyed listening to pure sinusoidal tones or square waves (I personally find them quite grating). A frequency counter is only good if you're counting something periodic, and not when something is very rapidly varying (as in music).