Some ultra cheap CMI8330 based cards do not have an onboard oscillator and use a clock signal provided by the motherboard, which may not be the right value for proper playback speed .

EDIT : See Re: Sound Blaster 16 Clones
Maybe adjusting your ISA BUS clock speed in BIOS will help .

If I reduce mp3 bitrate, there's a chance that the playback speed will improve?

Demetrio wrote on 2020-06-02, 22:43:

If I reduce mp3 bitrate, there's a chance that the playback speed will improve?

Improve how?

Audio accuracy, probably not
Audio loading time, skipping and buffering = better

I didn't see the edit so I asked (I'm not an expert).
I will try changing the clock speed in BIOS.
Thanks for the help.

Demetrio wrote on 2020-06-02, 22:43:

If I reduce mp3 bitrate, there's a chance that the playback speed will improve?

If CPU speed is insufficient, audio playback will skip, not slow down .

The slow playback speed is likely due to the card receiving an inappropriate clock signal, so lowering bitrate will not help at all .

Test it with an uncompressed wav file, you should have the same issue .

To fix this, you would probably need to modify the card by adding a proper clock oscillator (there used to be instructions on the Internet about how to do this, but I cannot locate them) . As mentioned before, if you are lucky, maybe changing the ISA bus clock speed in your BIOS will help (if you have such an option) .

darry wrote on 2020-06-02, 22:50:

If CPU speed is insufficient, audio playback will skip, not slow down .

Wrong. It will slow down with most MP3 decoders. Only a few skip actually, as they clear the buffer during playback (like Winamp e.g. does).

There's no way for a K5 with 90 MHz to play 320kbps MP3s. You need Pentium 75 speed for 128kbps joint stereo already.

K5 90Mhz is faster then P75..

No, it's not. MP3 uses floating point for decoding at which the K5 sucks. Also, you'd need that speed for a 128kbps joint stereo file, but he is trying to play 320kbps simple stereo files that need about 3 times as much horse power.

derSammler wrote on 2020-06-03, 11:10:

darry wrote on 2020-06-02, 22:50:

If CPU speed is insufficient, audio playback will skip, not slow down .

Wrong. It will slow down with most MP3 decoders. Only a few skip actually, as they clear the buffer during playback (like Winamp e.g. does).

That has not been my experience when attempting playback back in the day on a 486 with the multiple Windows players I tried at the time (among which Winplay3 and Winamp) . All of them skipped (presumably when the buffer underflowed) . I would imagine that slow playback could happen if the player has trouble even feeding rhe sound card fast enough from its buffer (if it has one), because the PC is really much too slow, but that would still give at least glitchy audio as samples would certainly be dropped unevenly, which is not what the OP described . Please provide examples of MP3 players that provide slower glitch-free playback on underpowered machines (I am of course not referring to the ones that let you adjust the playback speed intentionally, that is different).

Anyway, I suggested playing back a wave file to be sure it was a sound card clock issue and not related to MP3 decoding capability.

I remembered that the CMI8330 on the pc was an ISA card but I found out that is connected directly on the motherboard.

By the way, I will try playing a wav and see what happens.

Yeah, pretty sure you need a faster CPU. Try 128kbps files and see how they perform.

According to this blog, a K5 75Mhz can handle MP3 file decoding of an unspecified bitrate at 42% CPU usage .

http://bobblick.com/techref/projects/yammp3/yammp3.html

I played a wav in DSL: it seems to be better, maybe tiny slower but it is barely noticeable.

You can try older Winamp (2.x line) it has mp3 decoder which can be configured for older processors

Image is taken from this thread Lowest CPU to play MP3s?

zyga64 wrote on 2020-06-04, 06:27:

You can try older Winamp (2.x line) it has mp3 decoder which can be configured for older processors https://i.imgur.com/zkqNisq. […]
Show full quote

You can try older Winamp (2.x line) it has mp3 decoder which can be configured for older processors

Image is taken from this thread Lowest CPU to play MP3s?

Winplay3 is also worth a shot. It is very old and optimised for the machines of its era .

But if that wav file played slower than normal, it is still strange . Was the wav file the same as song as one of the MP3s you tried? It would make the comparison easier .

darry wrote on 2020-06-04, 06:33:

But if that wav file played slower than normal, it is still strange . Was the wav file the same as song as one of the MP3s you tried? It would make the comparison easier .

Yes, it was.
The wav has the same sample rate of the mp3 and a bitrate of 1411 kb/s.
Like the mp3s, I tested it by playing it with the same song on my phone at the same time: it seemed that it was tiny slower but maybe I perceived wrong.
I didn't tested a lot so I can be wrong.

I retested the wav: it is slower.
I played together the song slightly before on the pc and shortly after on my phone: the phone exceeded the pc.

The first reply of this thread told you exactly the problem so why waste so much time testing it? If a 68000 can play an MP3 (as long as it's the only thing it's doing), your CPU is capable.

pii_legacy wrote on 2020-06-04, 18:24:

The first reply of this thread told you exactly the problem so why waste so much time testing it? If a 68000 can play an MP3 (as long as it's the only thing it's doing), your CPU is capable.

Other posts said that it could be the CPU so I wanted to be sure.
The CMI8330 is connected directly to the motherboard so I don't know how to fix this; maybe I will buy another audio card.