My primary interest is radio communication, I was wondering if anyone out there had any circuits for a signal processor project that i could build and install inside my receiver, that would connect inline with the audio output path,
I know there are products avaliable, like the BHI DSP (NEDSP1061) noise elimination products but unfortunatly the £100 price tag is well out of reach.
I dont know much about the DSP that you have asked for but the ones from Texas instruments e.g C6713 has a lot of support.Try edaboard.com there is a whole section on DSP's there.
I dont know much about the DSP that you have asked for but the ones from Texas instruments e.g C6713 has a lot of support.Try edaboard.com there is a whole section on DSP's there.
Doesn't that count as software radio? Where all the demodulation and processing is done in a DSP instead of with oscillators, filters, and mixers?
If it is that is subject of quite a lot of research seeing as there any ADCs that are fast enough or have a high enough resolution to digitize high frequency, low level RF signals so all the signal processing can be done in a DSP. I am unsure if there exist DSPs that are capable of that task, but the big problem seems to be the ADC right now.
no, all i am looking for is a filter circuit thar goes inline with audio output, before the speaker or headphones, some type of low/high pass circuit? I built one as a teenager a long time ago! but have no recollection of the details and don't have the circuit now!
Simple filters remove the important consonants sounds of speech so that people talk like ducks quacking. Consonants have frequencies from 8kHz to 14kHz.
I know the DSP can select the normal vocal range, and then attempt to attenuate whats left, I thought that a analog circuit may be able to remove some of the background and leave inteligable speech.
I know you can use high/low pass audio filters for CW (morse code) but the tone is only within a tight audio frequency range,
If theres nothing out there, well you can't win them all.
Reducing noise by cutting high frequencies is a compromise.
If the noise is very bad then without cutting high frequencies the noise masks the speech.
If the high frequencies are cut then the consonants of speech are also cut and the intellegibility is poor.
FIG. 4 shows graphs of an unclipped voice signal (upper curve) and a clipped and band-filtered signal according to the circuit shown the FIG. 2 (lower curve). It has a band-width of 300 to 2050 Hertz and has been judged by a poll of listerners to provide a good degree of voice articulation combined with satisfactory speaker recognition.
Thanks Mike,
I am glad that Polycom agrees with me that a narrow bandwidth for speech sounds like hell, has poor intellegibility and is unnatural.
They even agree that ordinary 300Hz to 3.3kHz telephone bandwidth is poor.
You can download Filter Pro from TI which can help you design your own AF filters using opamps. Using a DSP is more difficult and not recommended if you don't already have programming experience with microcontrollers.