making audio bit by bit

[Simon Blackmore]

Sorry if this has been covered before but has anyone here explored  
making audio bit by bit

I am not really sure where to start. Say for example I wanted to make  
1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at  
all possible

It seems a bit crazy, but I am sure there is someone crazy enough to  
have explored it here.

Cheers

Simon




_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[unaudio]

That would result in silence

On Fri, Jul 4, 2008 at 2:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon




_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[Simon Blackmore]

Yeah apparently its an AC silence at DC-offset 0x1515. I am just curious how you implement such a sound or non sound. I was just looking into the possibilities of burning a visual image onto an audio CD in 0s and 1s. Apparently as this byte gets translated to EFM 00000000100000 on a CD there is a noticeable visual between that and data with more 1s in it. 

Cheers

Simon

On 4 Jul 2008, at 14:31, Vytautas Jancauskas wrote:

That would result in silence

On Fri, Jul 4, 2008 at 2:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon




_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[James Harkins-2]

On Jul 4, 2008, at 5:28 AM, simon blackmore wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples

with the following 16 bits 0010000000100000 in each sample. Is this at all possible

Not too hard...

The only tricky thing is that supercollider always processes sound data normalized to -1.0 .. 1.0, regardless of the bit depth of the hardware output or sound file. So if your input value is relative to 16 bits, you should divide by 32768 to get the corresponding floating-point value.

// as a server buffer

b = Buffer.alloc(s, 44100, 1, completionMessage: { |buf|

buf.fillMsg(0, 44100, 2r0010000000100000 / 32768)

});

// prove the values are OK

b.getn(0, 100, { |v| (v * 32768).postln });

// or, write directly to a soundfile from the language

f = SoundFile("sounds/bits.aiff")

.numChannels_(1)

.headerFormat_("AIFF")

.sampleFormat_("int16");

if(f.openWrite) {

d = FloatArray.fill(44100, 2r0010000000100000 / 32768);

f.writeData(d);

};

f.close;

// prove that it's good

f = SoundFile.openRead("sounds/bits.aiff");

if(f.isOpen) {

d = FloatArray.newClear(100);

f.readData(d);

(d * 32768).postln;

};

f.close;

hjh

: H. James Harkins

: jamshark70@...

: http://www.dewdrop-world.net

.::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..:

"Come said the Muse,

Sing me a song no poet has yet chanted,

Sing me the universal."  -- Whitman

[felix-38]

g = Signal.fill(44100,{ arg i;    your numbers here })

you can then save it to disk,

g.write(path)

load it to the server

b = Buffer.read(server,path)

and play it with a PlayBuf
of a SoundFile if its very large.

you can also open an audio file for writing and then write numbers directly to the file.




also aSignal.waveFill( function,

etc.


-cx

On Fri, Jul 4, 2008 at 11:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon

[Simon Blackmore]

Thanks James, Felix,

That seems much easier than I had thought.

Cheers

Simon

g = Signal.fill(44100,{ arg i;    your numbers here })

you can then save it to disk,

g.write(path)

load it to the server

b = Buffer.read(server,path)

and play it with a PlayBuf
of a SoundFile if its very large.

you can also open an audio file for writing and then write numbers directly to the file.




also aSignal.waveFill( function,

etc.


-cx

On Fri, Jul 4, 2008 at 11:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon

[Luke Selden]

Fredrik Olofsson made some Ugens that perform bitwise calculations,
which may be helpful in your explorations.  You may also want to look
at his chip emulation Ugens, because they allow/encourage bitwise
changes to their parameters.

http://www.fredrikolofsson.com/pages/code-sc.html


~luke

>
> On Fri, Jul 4, 2008 at 11:28 AM, simon blackmore <simon@...>
> wrote:
>>
>> Sorry if this has been covered before but has anyone here explored making
>> audio bit by bit
>>
>> I am not really sure where to start. Say for example I wanted to make 1
>> second or 44100 samples
>> with the following 16 bits 0010000000100000 in each sample. Is this at all
>> possible
>>
>> It seems a bit crazy, but I am sure there is someone crazy enough to have
>> explored it here.
>>
>> Cheers
>>
>> Simon
>>
>>
>>
>
>

_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[krgn]

I always wanted to look into doing this in bash with dd or and stdin, havn't really gotten around to it though :)

On Fri, Jul 4, 2008 at 2:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon




_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[James Harkins-2]

Just to be absolutely sure... I opened "bits.aiff" in a hex editor and the audio data are

0x202020202020202020...

And 0x2020 == 2r0010000000100000, so dividing by 32768 is right. There should be no fp rounding error because 32768 is a power of two.

hjh

On Jul 4, 2008, at 11:01 AM, James Harkins wrote:

// or, write directly to a soundfile from the language

f = SoundFile("sounds/bits.aiff")

.numChannels_(1)

.headerFormat_("AIFF")

.sampleFormat_("int16");

if(f.openWrite) {

d = FloatArray.fill(44100, 2r0010000000100000 / 32768);

f.writeData(d);

};

f.close;

: H. James Harkins

: jamshark70@...

: http://www.dewdrop-world.net

.::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..:

"Come said the Muse,

Sing me a song no poet has yet chanted,

Sing me the universal."  -- Whitman

[Simon Blackmore-2]

Thanks James, Felix and Luke and all,

That seems much easier than I had thought.

Cheers

Simon

g = Signal.fill(44100,{ arg i;    your numbers here })

you can then save it to disk, 

g.write(path)

load it to the server

b = Buffer.read(server,path)

and play it with a PlayBuf
of a SoundFile if its very large.

you can also open an audio file for writing and then write numbers directly to the file.




also aSignal.waveFill( function,

etc.


-cx

On Fri, Jul 4, 2008 at 11:28 AM, simon blackmore <simon@...> wrote:

Sorry if this has been covered before but has anyone here explored making audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1 second or 44100 samples
with the following 16 bits 0010000000100000 in each sample. Is this at all possible

It seems a bit crazy, but I am sure there is someone crazy enough to have explored it here.

Cheers

Simon

Fredrik Olofsson made some Ugens that perform bitwise calculations,
which may be helpful in your explorations.  You may also want to look
at his chip emulation Ugens, because they allow/encourage bitwise
changes to their parameters.

http://www.fredrikolofsson.com/pages/code-sc.html


~luke

On Fri, Jul 4, 2008 at 11:28 AM, simon blackmore <simon@...>

wrote:

Sorry if this has been covered before but has anyone here explored making

audio bit by bit

I am not really sure where to start. Say for example I wanted to make 1

second or 44100 samples

with the following 16 bits 0010000000100000 in each sample. Is this at all

possible

It seems a bit crazy, but I am sure there is someone crazy enough to have

explored it here.

Cheers

Simon

_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/

[Marinos Koutsomichalis]

Could you explain why do we have to divide it by 32768??  What does this number mean?

On 04 Ιουλ 2008, at 10:08 ΜΜ, James Harkins wrote:

Just to be absolutely sure... I opened "bits.aiff" in a hex editor and the audio data are

0x202020202020202020...

And 0x2020 == 2r0010000000100000, so dividing by 32768 is right. There should be no fp rounding error because 32768 is a power of two.

hjh

On Jul 4, 2008, at 11:01 AM, James Harkins wrote:

// or, write directly to a soundfile from the language

f = SoundFile("sounds/bits.aiff")

.numChannels_(1)

.headerFormat_("AIFF")

.sampleFormat_("int16");

if(f.openWrite) {

d = FloatArray.fill(44100, 2r0010000000100000 / 32768);

f.writeData(d);

};

f.close;

: H. James Harkins

: jamshark70@...

: http://www.dewdrop-world.net

.::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..:

"Come said the Muse,

Sing me a song no poet has yet chanted,

Sing me the universal."  -- Whitman

[unaudio]

This is the highest value a sample can take, dividing it by that scales it to -1.0, 1.0

On Mon, Jul 7, 2008 at 6:55 AM, Marinos Koutsomichalis <marinos@...> wrote:

Could you explain why do we have to divide it by 32768??  What does this number mean?

On 04 Ιουλ 2008, at 10:08 ΜΜ, James Harkins wrote:

Just to be absolutely sure... I opened "bits.aiff" in a hex editor and the audio data are

0x202020202020202020...

And 0x2020 == 2r0010000000100000, so dividing by 32768 is right. There should be no fp rounding error because 32768 is a power of two.

hjh

On Jul 4, 2008, at 11:01 AM, James Harkins wrote:

// or, write directly to a soundfile from the language

f = SoundFile("sounds/bits.aiff")

.numChannels_(1)

.headerFormat_("AIFF")

.sampleFormat_("int16");

if(f.openWrite) {

d = FloatArray.fill(44100, 2r0010000000100000 / 32768);

f.writeData(d);

};

f.close;

: H. James Harkins

: jamshark70@...

: http://www.dewdrop-world.net

.::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..:

"Come said the Muse,

Sing me a song no poet has yet chanted,

Sing me the universal."  -- Whitman

[Marinos Koutsomichalis]

Ok, so what we actually do is convert the values to decimal float point ones. right?

On 07 Ιουλ 2008, at 5:02 ΜΜ, Vytautas Jancauskas wrote:

This is the highest value a sample can take, dividing it by that scales it to -1.0, 1.0

On Mon, Jul 7, 2008 at 6:55 AM, Marinos Koutsomichalis <marinos@...> wrote:

Could you explain why do we have to divide it by 32768??  What does this number mean?

On 04 Ιουλ 2008, at 10:08 ΜΜ, James Harkins wrote:

Just to be absolutely sure... I opened "bits.aiff" in a hex editor and the audio data are

0x202020202020202020...

And 0x2020 == 2r0010000000100000, so dividing by 32768 is right. There should be no fp rounding error because 32768 is a power of two.

hjh

On Jul 4, 2008, at 11:01 AM, James Harkins wrote:

// or, write directly to a soundfile from the language

f = SoundFile("sounds/bits.aiff")

.numChannels_(1)

.headerFormat_("AIFF")

.sampleFormat_("int16");

if(f.openWrite) {

d = FloatArray.fill(44100, 2r0010000000100000 / 32768);

f.writeData(d);

};

f.close;

: H. James Harkins

: jamshark70@...

: http://www.dewdrop-world.net

.::!:.:.......:.::........:..!.::.::...:..:...:.:.:.:..:

"Come said the Muse,

Sing me a song no poet has yet chanted,

Sing me the universal."  -- Whitman

[James Harkins-2]

On Mon, Jul 7, 2008 at 9:55 AM, Marinos Koutsomichalis
<marinos@...> wrote:
> Could you explain why do we have to divide it by 32768??  What does this
> number mean?

Simon's initial email specified 16 bits of precision. In the integer
world, there are 2**16 possible integers that can be expressed using
16 bits, or 65536. Audio signals are signed values, so one bit is used
for the sign, leaving 15 bits to represent amplitude, or 32768.

So the floating point range -1.0 .. 1.0 gets scaled in the hardware
output to -32768 .. 32767 if the hardware out is configured for 16
bits. And, if you want to scale a known integer amplitude into -1.0 ..
1.0, you divide by the largest possible integer amplitude, which is
the absolute value of -32768.

32767 == 2r0111111111111111 or the largest positive signed integer
that will fit into 16 bits (0 is the sign bit)
-32768 == 2r1000000000000000 or the largest negative signed integer
that will fit into 16 bits

hjh

--
James Harkins /// dewdrop world
jamshark70@...
http://www.dewdrop-world.net

"Come said the Muse,
Sing me a song no poet has yet chanted,
Sing me the universal." -- Whitman

_______________________________________________
sc-users mailing list


info (subscribe and unsubscribe): http://swiki.hfbk-hamburg.de:8888/MusicTechnology/880
archive: http://www.listarc.bham.ac.uk/marchives/sc-users/
search: http://www.listarc.bham.ac.uk/lists/sc-users/search/