Discussion:
Inaudible Audio
(too old to reply)
Kinfe Tadesse
2006-05-01 15:04:43 UTC
Permalink
Hello,

After successfully transmitting audio to CAPI, what I hear on the phone is
terribly poor; in fact, it is inaudible. The input audio to CAPI is alaw
encoded, is bit-flipped and sampled at 8000Hz. What could be the possible
reason for such a degradation? What should I try in order to improve the
audio output?

I appreciate any help.

Best regards,

Kinfe T.
Kinfe Tadesse
2006-05-08 16:54:27 UTC
Permalink
Oh, I solved the problem myself. I increased the maximum Data length value
(set in CAPI_REGISTER) to some higher value (8192) and I could listen to the
audio now (though it is not a pleasant one!)

KT
Post by Kinfe Tadesse
Hello,
After successfully transmitting audio to CAPI, what I hear on the phone is
terribly poor; in fact, it is inaudible. The input audio to CAPI is alaw
encoded, is bit-flipped and sampled at 8000Hz. What could be the possible
reason for such a degradation? What should I try in order to improve the
audio output?
I appreciate any help.
Best regards,
Kinfe T.
Werner Henze
2006-05-09 19:16:02 UTC
Permalink
Hi!
Post by Kinfe Tadesse
Oh, I solved the problem myself. I increased the maximum Data length value
(set in CAPI_REGISTER) to some higher value (8192) and I could listen to the
audio now (though it is not a pleasant one!)
So why isn't it pleasant? What kind of problem do you hear?
How do you do your sending of data? How do you keep it running?
Did you implement my guideline from your thread "Problem with DATA_B3_REQ"?

Ciao,
Werner...
--
PGP 8 available
http://home.arcor.de/werner.henze
Kinfe Tadesse
2006-05-10 15:35:37 UTC
Permalink
Dear Sir,
Post by Werner Henze
Post by Kinfe Tadesse
Oh, I solved the problem myself. I increased the maximum Data length value
(set in CAPI_REGISTER) to some higher value (8192) and I could listen to the
audio now (though it is not a pleasant one!)
So why isn't it pleasant? What kind of problem do you hear?
It is audible and I can hear what is being said because I know the prompts
it is playing. Considering a real life scenario, it is very hard for someone
to understand what is being said.
Post by Werner Henze
How do you do your sending of data? How do you keep it running?
By the way, the prompts are being played by a TTS. I take audio data from
the TTS and stream it to CAPI. The problem was the TTS streams buffers with
length 8192 at a time. When I set the Data length to that value it worked.
However, this resulted in another disastrous problem. With this data length
set, DATA_B3_IND can never be detected (a question in another thread). It
took me considerable time to identify this problem. Is there any maximum
size for buffer length (the value set in CAPI_REGISTER)?
Post by Werner Henze
Did you implement my guideline from your thread "Problem with
DATA_B3_REQ"?
I tried to consider your guidelines. But I failed to implement the "bucket
brigade" comment you suggested;i.e., sending 2 or more data requests before
getting a Data confirmation. If I could implement this, then the data size
problem will also be solved. Please advise me how to implement this!

Best regards,

KT.
Tobias Erichsen
2006-05-10 16:05:00 UTC
Permalink
Post by Kinfe Tadesse
Post by Werner Henze
How do you do your sending of data? How do you keep it running?
By the way, the prompts are being played by a TTS. I take audio data from
the TTS and stream it to CAPI. The problem was the TTS streams buffers
with length 8192 at a time. When I set the Data length to that value it
worked. However, this resulted in another disastrous problem. With this
data length set, DATA_B3_IND can never be detected (a question in another
thread). It took me considerable time to identify this problem. Is there
any maximum size for buffer length (the value set in CAPI_REGISTER)?
I really don't understand why you don't take any of the suggestions
quite a few people on this newsgroup have given to you, with any
more seriousness. Those people who have responded to you
are working with CAPI for years and know what they are talking
about.

You don't seem to have any scope about state-machines, event-driven
programming and software-engineering in general. I have before
suggested to take a look at the stuff that the publicVoiceXML
people are doing with CAPI - I don't know how good it is,
but even if it sucks bad, it will be better than the stuff you have so far
shown to this newsgroup - sorry to be this frank, but I think you
don't get it any other way!

There are also quite a few other open-source things out there working with
CAPI (both on linux & win32) which you should study carefully and twiddle
around with for quite a while before you are ready to roll your own stuff.

Tobias

Loading...