VOGONS

Common searches


Question about Wave files

Topic actions

Reply 20 of 38, by m1so

User metadata
Rank Member
Rank
Member

An interesting bit from the history of digital sound, namely, the famous Linn LM-1 drum machine:

http://www.synthmuseum.com/linn/linlm101.html

Roger Linn wrote:

"'I believe the LM-1 sounded better because I didn't incorporate strict textbook digital sampling theory. By the book, I should have filtered out any playback frequencies above the Nyquist frequency, which is a little less that one-half of the sampling frequency. I used a sampling rate of around 27kHz. However, filtering on playback would have made some of the drums sound pretty dull. Instead, I let some of the frequencies above that point get through, because the results - which can get distorted - sounded like the sizzle of drums anyway. Thanks to that decision, the LM-1 sounded better than some drum machines with the same sampling rate, because it had the highs. In a sense, I'm thankful that I wasn't very good at the engineering.'"

Reply 22 of 38, by Stojke

User metadata
Rank l33t
Rank
l33t

I am very interested but im busy at the moment 😀
I read your first post and i want to read the others when i have full attention to it.

Im currently recording some MIDIs from my SW1000XG, while I am working on a project for one person.

http://www.youtube.com/watch?v=AXO0lBPmz30
http://www.youtube.com/watch?v=53iMoO0M86o

Note | LLSID | "Big boobs are important!"

Reply 23 of 38, by VileR

User metadata
Rank l33t
Rank
l33t
m1so wrote:

Everyone lost interest 🙁 ?

Nope- this is actually quite informative. I know enough about sampling theory and have produced my own music before, but I'm a "go with what you know" guy and all this bitcrusher stuff is news to me.
Though I'm a little perplexed by Mr Linn's explanation of the Nyquist frequency being "a little less that one-half of the sampling frequency". As far as I know it's exactly one half... unless he means it in a practical sense of taking precautions.

[ WEB ] - [ BLOG ] - [ TUBE ] - [ CODE ]

Reply 24 of 38, by m1so

User metadata
Rank Member
Rank
Member

Probably because lowpass filters have a certain rolloff, so in practice yes, it would be a little below half the sampling rate.

Some more samples, this time done with Decimort, the "Amiga 1200 filter" files I lowpass filtered in Audacity using a 5000 Hz / 6 dB/oct setting as Wikipedia says.

http://www.fileswap.com/dl/GjzX4t1T8i/
http://www.fileswap.com/dl/6Tekv4rg8C/
http://www.fileswap.com/dl/Gq5406eucs/
http://www.fileswap.com/dl/biSrfuPaOV/
http://www.fileswap.com/dl/TVr7btSd2N/
http://www.fileswap.com/dl/d3j6UYHIkL/
http://www.fileswap.com/dl/cjClRvo18G/
http://www.fileswap.com/dl/G1bA8KTB/
http://www.fileswap.com/dl/4p0gvIH3Q/

16 Khz can sound surprisingly good.

As for pure voices, I cannot find my only audiobook USB key right now, but honestly, unless you go below 6 bit the noise should not really be such a problem.

Reply 25 of 38, by Jepael

User metadata
Rank Oldbie
Rank
Oldbie
m1so wrote:

Note that this is the "worst case scenario", with both AD and DA aliasing at 100 percent (adjustable in the settings). Most old computers would have a filter, but a relatively gentle one, not the strict one that most new computers have, and that is why for example Amiga mod music sounds so awesome despite most samples being around 16 Khz.

Not only about filter, but the difference is more in the fundamental difference between how MOD music was played on Amiga natively versus on PC.

Amiga has four separate DAC channels and each DAC can update the sample after update period value that is based on 3.58 MHz clock, versus software mixing to one single audio stream with fixed sampling rate of say 44.1kHz.

Software mixing on PC is usually done with just by calculating which instrument sample would be playing when output sample needs to be generated, and that causes uneven skipping or doubling of instrument samples when the software tries to achieve the required average instrument playback speed. On an Amiga, it never skips or doubles samples, it plays each sample an equal amount of time ticks, the period value. This is the main source of bad sound. A Gravis Ultrasound can playback MODs a bit better as it uses hardware interpolation and mixing of channels, but just using linear interpolation while software mixing won't save from this issue even though it sounds just somewhat better.

As far as I know, only one MOD player (UADE) tries to emulate the Amiga DAC update resolution correctly with actual step responses of the system. It's almost the same thing than actually generating four channel DAC output at 3.58MHz and filtering and downsampling that to 44.1kHz for playing.

Reply 26 of 38, by m1so

User metadata
Rank Member
Rank
Member

Jepael, you are right. I guess my samples are more like early Sound Blaster sound, which was quite "aliasey" (Amiga has aliasing, but a bit less harsh sounding). Try turning off interpolation and micro volume ramping in your mod player plugin and you'll get that "soundblastery" aliasing sound even now. What is funny is that OPL2/OPL3 FM music was very clean sounding even on Sound Blaster 1.0 as the DAC on the OPL chips use a 49716 Hz sample rate and 13-bit floating point PCM bit depth.

I promised voice samples, so here they are:

http://www.fileswap.com/dl/gN4WtGOjfJ/
http://www.fileswap.com/dl/8Y1OobRnn/
http://www.fileswap.com/dl/K6RlfwXFmj/
http://www.fileswap.com/dl/YtHaYyabXu/

At around 8 Khz the aliasing becomes nastier than the filter, so for 8 Khz and lower full filtering is better. I also added a little "bonus", a 1-bit DPCM (Nintendo NES, internally decoded to 6-bit) sample at the highest sample rate possible on the NES made with the demo of Chipcrusher. Sounds quite okay considering DPCM at cca 33 Khz has half the bitrate of a telephone call and without use of any psychoacoustic algorithms.

At 8-bit without dithering, the quantization noise modulates the sound, so there is little background noise, only little "swishing noise" in the speech. 1-bit DPCM is noisy as hell, but still good considering the extreme bit reduction.

Sorry for the samples being stereo, sourced from an mp3 audiobook and not English, but I have no other high quality speech samples to downsample.

Reply 27 of 38, by mr_bigmouth_502

User metadata
Rank Oldbie
Rank
Oldbie

I wonder, what was the audio hardware like in early 90s Macintosh computers? I was reading here http://en.wikipedia.org/wiki/Broken_%28EP%29 that Trent Reznor of Nine Inch Nails created all of the guitar parts for the Broken EP by playing through a Zoom pedal and processing the sound in Digidesign TurboSynth. I love how the guitars sound on this EP, and I wonder how someone could "replicate" this effect.

Here are some examples:

Last https://www.youtube.com/watch?v=MeoyOIRIeHs
Wish https://www.youtube.com/watch?v=sZ4gpR579Ys
Gave Up https://www.youtube.com/watch?v=9TiKC3ugR3s

Last edited by mr_bigmouth_502 on 2013-09-29, 02:10. Edited 1 time in total.

Reply 28 of 38, by Jepael

User metadata
Rank Oldbie
Rank
Oldbie
m1so wrote:

What is funny is that OPL2/OPL3 FM music was very clean sounding even on Sound Blaster 1.0 as the DAC on the OPL chips use a 49716 Hz sample rate and 13-bit floating point PCM bit depth.

I should know 😀 The DAC has only 10-bit resolution but 16-bit range. The 3 bits select the range. The OPL3 chip has pure 16-bit output, but it gets converted to 10-bit resolution with 16-bit range in the DAC so it is similar. For comparison, the output of a single FM operator has 12 bit resolution and 13 bit range (-4085 to 4084).

Also the sound reproduction mechanism in OPL is also just like software mixing at fixed frequency. There is one wave of sine with 1024 samples (actually one quarter stored) and this 1024-entry lookup table is indexed with 10 most significant bits of a 19-bit phase accumulator.

There is some aliasing apparent in some sounds, but not much. It might be from digital steps when for example envelope generator changes amplitude in steps few hundred times per second while generating the sine. Also not many OPL3 cards have implemented any kind of filter after OPL3 DAC. A real Adlib had a four-pole Butterworth filter at 15.4 kHz. I really should get one.

Reply 29 of 38, by HunterZ

User metadata
Rank l33t++
Rank
l33t++
leileilol wrote:

CoolEdit 95/96 can do that. Quake is the earliest game to make use of it (to my knowledge) and it is what id Software used.

On PC, ADPCM was really used far later in the 90s when things were phasing away from CD audio for music and MP3 licensing was still too expensive. SiN and GTA2 are nice ADPCM game examples. Quake3 has bits of ADPCM decoding in the code, but never was actually used.

Of course there's also to mention the support for compressed sounds in the earlier SB cards... though i can't name a bunch of games that used it off the top of my head.

Interplay used various pre-MP3 compression algorithms for the soundtrack files in their Win9x-era games like Fallout 1 & 2 and the various Baldur's Gate engine games (BG1&2, Planescape: Torment, Icewind Dale 1 & 2, etc.).

My impression is that part of the problem with MP3 adoption for game soundtracks was that it was CPU-intensive when it first popped up. I had a 486DX4-120 at the time, for example, and had to listen to MP3s at a lower quality than what the source files were recorded at. Of course, Pentiums were probably better at it because they were so much better at floating-point math.

Reply 30 of 38, by m1so

User metadata
Rank Member
Rank
Member

mr_bigmouth_502, as far as I know, early Macs had similiar built in audio hardware to the early Sound Blaster cards, minus the OPL2/3 part. I think most musicians who wanted to make music on the Mac at that time bought an addon soundcard/module that enabled studio quality sound, but this is industrial music and this is Trent Reznor, I think it would not be surprising for him to use the native sampling hardware to add some dirt to his sound.

On a side note, 1-bit DPCM at 33.144 Khz sounds quite OK, even with music, but at samplerates below 8 Khz it creates the most harsh, horrifying lo-fi mess ever imaginable. At 4 Khz, even speech is literally impossible to understand. Just look in this video starting on around 1:45. That is 33 Khz DPCM and it sounds quite good for such a low bitrate. Then listen to the guy switching from NES quality F (33 Khz, highest) to NES quality 0 (4 Khz, worst) and at 2:05 you have a sound that can only be described as the screams of damned dial up modems being tortured in hell.

Reply 32 of 38, by mr_bigmouth_502

User metadata
Rank Oldbie
Rank
Oldbie
m1so wrote:

mr_bigmouth_502, as far as I know, early Macs had similiar built in audio hardware to the early Sound Blaster cards, minus the OPL2/3 part. I think most musicians who wanted to make music on the Mac at that time bought an addon soundcard/module that enabled studio quality sound, but this is industrial music and this is Trent Reznor, I think it would not be surprising for him to use the native sampling hardware to add some dirt to his sound.

I know he used an early version of ProTools as well, but I wouldn't be surprised if he ignored the soundcard that came with it in favor of the onboard audio. Would it have been possible to switch between two audio devices on a Mac of that era?

On a related note, I know that Skinny Puppy did a lot of stuff using the Ensoniq Mirage sampler and the Ensoniq ESQ-1 synth, both of which feature the same sound chip found in the Apple IIGS. Has anyone ever rigged up a GS to use it for really gritty sounding sampling or synthesis? I know it lacks the analog Curtis filters, but it sure as hell would be cool to take the machine I have and use it to make oldschool industrial music.

Reply 33 of 38, by m1so

User metadata
Rank Member
Rank
Member
VileRancour wrote:

Video link missing?

Ooops 😀 here it is http://www.youtube.com/watch?v=7XaRds4e3eo .

Anyways, there is another funny thing about digital sound. The success of mp3 has more to do with just how flawed our ears are than with the quality of it. I am not speaking about the psychological "eh, 128 kbps is enough" thing, but about the fact that the signal to noise ratio of mp3 is about the same as the SNR of 2-4 bit ADPCM. Similiar to ADPCM, it has a giant dynamic range, yet small signal to noise ratio (ADPCM usually has a DR of 96 dB and SNR of around 24-40 dB while mp3 has a DR of around 125 dB and a SNR of 25 dB). What makes mp3 and other psychoacoustic formats different is that they mask the noise using psychoacoustic methods. The mathematical "bit depth" of 128 kbps mp3 is just under 3 bits and that of 320 kbps mp3 is just over 7 bits. A really bad encoder makes the noise just above the hearing threhold, while a good encoder like LAME masks it so well that it is not just hidden, but literally impossible to hear at high bitrates by human ears.

So yeah, mp3 is not actually "16-bit". The internal data is 32-bit floating point, but the number of bits per sample is anywhere between 1 to 7. Many people think ADPCM is some kind of an atrocity because they see "4-bit" and think "zomg Atari 2600 sound". Not many people know that their 128 kbps "CD quality" mp3s have a bit depth of 2.9 bit/sample.

ADPCM is usually inferior to mp3 because it has no psychoacoustic masking to hide the noise. There are however many exceptions. There are problem samples like the infamous castanets.wv or the beginning of Fatboy Slim's song Kalifornia that can be easily ABXed from the original even at 320 kbps. ADPCM adds a little noise in those cases, but otherwise handles them correctly because it is a purely "mathematic" compression system, while the psychoacoustic model of mp3 or aac just collapses on these problem samples.

EDIT - I forgot to say, those bittdepths are for mono mp3s. In stereo mp3s they are half that.

Reply 34 of 38, by m1so

User metadata
Rank Member
Rank
Member

The voice clip I posted before, in glorious 4.18 Khz 1-bit DPCM 🤣 http://www.fileswap.com/dl/akoDRBZ0dY/ .

Honestly, I don't think any NES game even used a sample rate below 16 Khz with DPCM. At 4.18 Khz, the sound basically becomes a hideously aliased, stairstepped triangle wave.

Reply 35 of 38, by Stiletto

User metadata
Rank l33t++
Rank
l33t++
mr_bigmouth_502 wrote:

On a related note, I know that Skinny Puppy did a lot of stuff using the Ensoniq Mirage sampler and the Ensoniq ESQ-1 synth, both of which feature the same sound chip found in the Apple IIGS. Has anyone ever rigged up a GS to use it for really gritty sounding sampling or synthesis? I know it lacks the analog Curtis filters, but it sure as hell would be cool to take the machine I have and use it to make oldschool industrial music.

I haven't but I know the people currently working on emulating the Ensoniq ESQ-1 and Mirage (in a source-available emulator called MESS) have:
http://git.redump.net/mame/tree/src/mess/drivers/esq1.c
http://git.redump.net/mame/tree/src/mess/drivers/mirage.c

It's still early days but some links you'll find interesting:
http://rbelmont.mameworld.info/?p=754
http://rbelmont.mameworld.info/?p=844
http://rbelmont.mameworld.info/?p=853

"I see a little silhouette-o of a man, Scaramouche, Scaramouche, will you
do the Fandango!" - Queen

Stiletto

Reply 36 of 38, by mr_bigmouth_502

User metadata
Rank Oldbie
Rank
Oldbie
Stiletto wrote:
I haven't but I know the people currently working on emulating the Ensoniq ESQ-1 and Mirage (in a source-available emulator call […]
Show full quote
mr_bigmouth_502 wrote:

On a related note, I know that Skinny Puppy did a lot of stuff using the Ensoniq Mirage sampler and the Ensoniq ESQ-1 synth, both of which feature the same sound chip found in the Apple IIGS. Has anyone ever rigged up a GS to use it for really gritty sounding sampling or synthesis? I know it lacks the analog Curtis filters, but it sure as hell would be cool to take the machine I have and use it to make oldschool industrial music.

I haven't but I know the people currently working on emulating the Ensoniq ESQ-1 and Mirage (in a source-available emulator called MESS) have:
http://git.redump.net/mame/tree/src/mess/drivers/esq1.c
http://git.redump.net/mame/tree/src/mess/drivers/mirage.c

It's still early days but some links you'll find interesting:
http://rbelmont.mameworld.info/?p=754
http://rbelmont.mameworld.info/?p=844
http://rbelmont.mameworld.info/?p=853

I actually own a GS, but I need to dig it out of storage. 🤣 That's what I meant when I was referring to "my machine".

Cool links anyhow. 😁 I'm actually rather surprised that the MESS people have gone as far as attempting to emulate musical keyboards, and not just consoles and computers.

Reply 37 of 38, by bloodbat

User metadata
Rank Oldbie
Rank
Oldbie

If I have some spare time...I'll give it a go.

mr_bigmouth_502 wrote:
I wonder, what was the audio hardware like in early 90s Macintosh computers? I was reading here http://en.wikipedia.org/wiki/Bro […]
Show full quote

I wonder, what was the audio hardware like in early 90s Macintosh computers? I was reading here http://en.wikipedia.org/wiki/Broken_%28EP%29 that Trent Reznor of Nine Inch Nails created all of the guitar parts for the Broken EP by playing through a Zoom pedal and processing the sound in Digidesign TurboSynth. I love how the guitars sound on this EP, and I wonder how someone could "replicate" this effect.

Here are some examples:

Last https://www.youtube.com/watch?v=MeoyOIRIeHs
Wish https://www.youtube.com/watch?v=sZ4gpR579Ys
Gave Up https://www.youtube.com/watch?v=9TiKC3ugR3s