USR Sportster Voice Sound File Format

• Messages sorted by:[ date ][ thread ][ subject ][ author ]

-----BEGIN PGP SIGNED MESSAGE-----

Hi Alain!

sorry for answering so late.

> From: [Alain LESN_] <alesne@belouga.cetaces.loc>
> Subject: Re: USR Sportster Voice Sound File Format
> 
> My (french) USR Sportster 28800 Vi works very well with vgetty since I
> have apply this patch (and some others..., but for others reasons than GSM
> compression) in 0.7.2 vgetty release : 
> 
> ----- cut here -----
> --- gsm.h.ORG Mon Jun 16 21:41:13 1997
> +++ gsm.h  Sun Jun 15 18:48:38 1997
> @@ -41,7 +41,7 @@
> typedef unsigned char  gsm_byte;
> typedef gsm_byte  gsm_frame[33]; /* 33 * 8 bits */
> 
> -#define GSM_MAGIC d    /* 13 kbit/s RPE-LTP */
> +#define GSM_MAGIC 0    /* 13 kbit/s RPE-LTP */
> 
> #define GSM_PATCHLEVEL 7
> #define GSM_MINOR 0
> ----- cut here -----

I tried to implement this. I changed the gsm library to accept the
magic byte 0 as a valid identifier. But I didn't change it for the
creation of gsm files. So the question is, if your Sportster accepts
data with a magic byte of 0x0d as valid or not. Otherwise we need
some more tricks, I'm afraid.

> Indeed, it's works for me (for my modem, whitch is detected as a "USR
> Courier V.34(+)"), but this patch is not applicable to ALL USR Vi, because
> USRobotics says (in a file that i've downloaded on their BBS): 

Are you talking about the mgetty detection or the vgetty one?

> "in the voice product based on the Sportster 28800 Vi architecture, the
> head of the GSM frame is set to zero", but for "the future Sportster 28800
> Voice architecture, the format is the same as the Sportster Vi with the
> exception of the head of the GSM frame (...) the head of the GSM frame is
> set with the value 13 (or D in hex) instead of zero in the Sportster 28800
> Vi GSM frame." 

That's a very interesting information. Maybe I should include an
option variable to the array in detect.c, so that we can set
individual options for different ATI codes, that are currently mapped
to one driver. This way we can differentiate between modems of one
manufacturer. Maybe this also helps for the problems with the
different behaviour for some Rockwell modems. Any comments?

> As this modem is wrongly detected as a Courrier (ATI0 --> 3366) but is in
> fact a Sportster (ATI3 --> USRobotics Sportster Vi 33600 Faxmodem), I
> don't know if it is easy for Marc to modify source code whithout breaking
> others modems...

It is detected as a USR modem. Look at detect.c:

  {ati, "3366",   NULL, &US_Robotics},

I think you looked at the mgetty detection. And apart from that it is
now easy to things like this, because the detection is now much more
flexible.

> For information: I have encountered some other problems with this modem.

If you have a patch to solve them, please mail it to me.

> And some others, like DLE q (SILENCE_DETECTED) when there is
> NO_VOICE_ENERGY (DLE s). This is very annoying with (cheaper) american
> modems that don't send calling tone when they dialout. Thus, I have alter
> the Marc's code to always trying data connection after some silence...

Did you check the rec_min_rec_len option?

> Hope this help... and sorry again for my english

Of course!

Bye,
Marc
_______________________________________________________________________________
Email: Marc.Eberhard@Poseidon.ThPhy.Uni-Duesseldorf.DE
WWW: http://Poseidon.ThPhy.Uni-Duesseldorf.DE/~marc/
PGP: http://Poseidon.ThPhy.Uni-Duesseldorf.DE/~marc/Marc_PGP.key

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3ia
Charset: noconv

iQCVAwUBNOKIiTR1rYTfG0TdAQHrYgQAvFbQOYntjQHMxYbB6X1HElzDPZdo0chT
UAq84BkGdWsazI4UD81iY5XgC0OGTjTPEGMaxqB/YuewpCKGMI7idvePy4lifgm0
BiRw6KBSZ+pV265UOWW/PI84gRIb2xWqH09AuAl3nKjExfvpIjgqYazouMo4cJ2U
eAUhXTSq8sw=
=VlGb
-----END PGP SIGNATURE-----
.