Help dumping tapes for MAME

[Count9929A]

Hello,

 

 I am sampling my TI99 tapes, for preserving and for using them in MAME. I have few experience and no special equipment (see below). Nevertheless I had zero problems in sampling original (pre-recorded) tapes and have MAME to read the obtained WAVs.

 

The story, unfortunately, is different for tapes recorded from my TI: MAME usually detects some error in data, or less frequently returns the NO DATA FOUND message. However, the same WAVs can be read correctly by the real TI99, connecting the white jack of the standard tape connector into the phones plug in my MacBook Pro. Therefore it seems that MAME is more sensitive to errors than the real TI. Do you have any suggestion to improve the recording and sampling process, in order to create WAVs readable from MAME from self-recorded tapes?

 

My sampling procedure:

1) record from the TI99 to tape; I tried to two different recorders (TI Program Recorded and a Philips model) nothing changed. Tried with recording volume at mid level

2) sampling of the tape in a Mac, connecting the line-out from the tape in the audio input jack port; I tried several setting for the volume level in the Mac port, no success; I used the Quicktime Player software to do the sampling, that saves in AIFF

3) conversion from AIFF to WAV made with VLC

 

 

[+mizapf]

I had some success with the tool Audacity. You may try to increase the volume, filter out noise etc. If you post a wav, I may see what I can do with it.

[Count9929A]

26 minutes ago, mizapf said:

I had some success with the tool Audacity. You may try to increase the volume, filter out noise etc. If you post a wav, I may see what I can do with it.

Thank mizapf. I'll try Audacity. Meanwhile, I post 2 short WAV. They are 2 different sampling of the same TI BASIC program. Each one was sampled with a different level of input volume (on the Mac).

EXP-FOR-NT_iMac-vol-50_PH-REC-vol-4.wav EXP-FOR-NT_iMac-vol-75_PH-REC-vol-4.wav

[HOME AUTOMATION]

I noticed that there is little to no modulation on the right channel.
Most/all laptop jacks are STERIO.
TI plugs/jacks are MONO.
The compatibility of MONO plugs to STERIO jacks is unpredictable.
How MAME responds to a STERIO .wav file is questionable...
These issues may be a factor.

[Count9929A]

1 hour ago, HOME AUTOMATION said:

I noticed that there is little to no modulation on the right channel.
Most/all laptop jacks are STERIO.
TI plugs/jacks are MONO.
The compatibility of MONO plugs to STERIO jacks is unpredictable.
How MAME responds to a STERIO .wav file is questionable...
These issues may be a factor.

that's right, I used a Y cable, with one of the 2 mono input in the recorder jack, and the other left unconnected. I imagine that that's not the recommended solution. However, I did exactly the same with pre-recorded tapes, and they are read correctly.

Meanwhile, I also made some tests with Audacity. It's defintely a better tool than what I was using earlier, so thanks a lot mizapf. With audacity I can sample as MONO. Unfortunately, even with Audacity and recording in MONO, MAME detects errors. I'll try again, playing with the settings, in the next days.

By the way, Audacity allows to choose many encodings for the WAV. Up to now I found out that 16 bit PCM and 32 bit PCM are recognized by MAME, while 24 bit PCM is not. @mizapf: what encodings are recognized by MAME, and which are better to use?

 

thanks to both @mizapf and @HOME AUTOMATION for the help!

[+mizapf]

One tip for checking the tape loading in MAME:

 

- Start MAME with debugger ("-debug")

- In the debugger, press CTRL-M, then select "TMS9929A VDP ':vdp' vram space memory"

- Enter 0680 in the text field left of it, press Enter

- Go to the debug main window, press F5 to start; watch the memory view while trying to load a tape

- You should see incoming data starting at 0700. If you don't see new data flowing into memory after the intro tone, you don't need to wait until the end; loading has already failed.

[SteveB]

The vol50 file was unreadable for me, but the vol75 worked with the Windows-Program TAPE994A version 3.1 after converting it to 8bit WAV PCM mono using Audacity (use "other uncompressed format" to get more options than in "Export to WAV"). Use Track/Mix to covert to mono before exporting. 

 

The TAPE994A has the advantage to process a whole tape with multiple programs at once. So if you have access to a Windows machine, this might work better than direct MAME one-by-one import. 

 

EXP-FOR-NT_iMac-vol-75_PH-REC-vol-4_mono.wav

test.dsk

count01

[HOME AUTOMATION]

Since there is some noise on the right channel, it's probably better to drop it all-together...

 

 

 

 

       P.S. I would like to take this opportunity to thank Google spell-injector...

              ...As I've always preferred to listen in STERIO(sic).

[+mizapf]

I think the level must be above ±0.5 to be recognized. I managed it with the vol50 file, but MAME still refuses to load it, although from its waveform, it looks quite nice already.

 

Too bad I never had a closer look at the tape support in MAME.

[Count9929A]

Hi mizapf, thanks for your efforts. Maybe I can try to make a sample with a different volume level. 50vol means the input volume of the audio port in the Mac was at 50%. Should I try with a lower value?

Which WAV encoding is better? Steve B suggests 8bit WAV PCM, while the sample that I posted was 16 bit PCM

[HOME AUTOMATION]

I notice there is a lot of audible noise leading up to the file output.

A lower level on the record input, concurrent with a higher level on the playback output, should result in less noise.

 

I can hear frequency instability in the lead tone... I wouldn't expect a successful load that way. This is probably due to stretching in the main drive belt. Cheap tapes can have a lot of drag.

 

The observation that commercial tapes load, while self-made tapes fail, suggests that the original record deck's tracking was misaligned. Readjusting the playback deck's alignment to match, could make a huge difference in high-end reproduction(important for data).

 

I imagine that most modern commercial soundcards are optimized for 16-Bit D/A.
Speakerphone MODEMs are better at 8-Bit D/A.

 

The TI only waits about 25 seconds for the data to start. You have about 15 seconds of dead space before the data starts. Shaving some of that off might provide a bit more timing leeway for the emulated tape positioning operations.

[+OLD CS1]

10 minutes ago, HOME AUTOMATION said:

I can hear frequency instability in the lead tone... I wouldn't expect a successful load that way. This is probably due to stretching in the main drive belt. Cheap tapes can have a lot of drag.

On a real system, this surprisingly does not have much impact.  I have had the sync header fade out almost to nothing and the TI will still catch it.

[SteveB]

9 hours ago, Count9929A said:

Which WAV encoding is better? Steve B suggests 8bit WAV PCM, while the sample that I posted was 16 bit PCM

This is a requirement of TAPE994 and might be different for MAME. 8 bit is actually a bad quality, but easier to analyze and half the size to store. 

[+mizapf]

MAME uses 16bit signed.

[Count9929A]

I made other attempts. No success but I found something strange. I recorded the TI BASIC program at low volume.

Then I sampled it 4 times, keeping the volume on the tape player low, but with different levels of the input volume on the Mac

1. sample at 25% of input volume: DATA ERROR on MAME, NO DATA FOUND on TI99
2. sample at 50% of input volume: DATA ERROR on MAME, NO DATA FOUND on TI99
3. sample at 75% of input volume: DATA ERROR on MAME, NO DATA FOUND on TI99
4. sample at 100% of input volume: NO DATA FOUND on MAME, OK on TI99

It's strange that MAME gives error where the real TI gives NO DATA, and NO DATA where the TI reads the sample. It's a kind of reverse behavior.

I am also thinking, given that  have tried using 2 different recorders, that the problem could be that my TI plays the recording sound in a bad way. So even if my TI can read back the sample, other can not. So I would be interested to know if some other real TI has success in reading the samples obtained from my recordings. I attach those at 100% and at 75% of input value. Could someone kindly check with his real TI?

 

PS: this time I did the sampling with Audacity, in MONO recording

S_iMac-vol-75.wav S_iMac-vol-100.wav