online now

We have 99 guests and no members online


Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1

TOPIC: Best possible sound quality

Best possible sound quality 3 years 4 months ago #516

I'm new to GinSing/Babblebot/SoundGin. I have experimented a bit with SpeakJet, but, in short words, the sound quality of synthesized sounds is by far not acceptable (speech is ok). I want to relax when I hear the sounds.

My project is an experiment mixing synthesized sound: 2 components of (preferably triangle) sound in the range of 250 Hz to 1 kHz with different volumes, and sometimes speech messages. I do not use Arduino but design an own PC-board and program the controller in plain old simple C.
As only about 20 different words plus digits are to be distinguished, synthesized speech will be ok.

Most of my questions are answered by surfing after registration, one remains:
Please give me some hints how to get best possible sound quality out of BabbleBot/GinSing under the following restrictions:
- Space on the board is very (VERY) restricted,
- I have only one supply voltage of 5 or 6 Volt, (can of course be regulated down)
- "normally" I use an earphone compareable to those used with cell phones, usage of a loudspeaker would also allow an extra amplifier.
- I'm by far NOT familiar with analog audio electronics.
Experiments with SpeakJet and an LM386-based amplifier were frustrating: The amplifier generates a lot of noise by itself, the double RC filters yielded speech unuseable or (with smaller Cs) had no effect.
I hope, the higher sampling frequency of GinSing will allow to produce both: understandeable speech and relaxing sounds.
The sound quality in the file iamginsing.mp3 between seconds 16..26 is impressive and would make my wishes become true - has any special filter/amplifier been used to produce this file?

Is BabbleBot/GinSing available in an SOC package? That would help a lot.

Thank you for your efforts so far, sincerely yours
The administrator has disabled public write access.

Re: Best possible sound quality 3 years 4 months ago #517

HI Helmut, thanks for the questions.

The output of the Babblebot IC is PWM, which means that a simple passive filter (as in the stage preceding the LM386 in the circuit diagram is the lowest noise output. Basically the chip outputs a digital pulse stream that varies in duration per cycle based on the output amplitude, but does so at a fixed frequency. If you were to connect this to a load directly, such as a small high impedance speaker, the speaker itself would filter out the high frequency modulation and produce an audible waveform. HOWEVER, the chip cannot take much current load ( 50 mA max), so on a 5V output I = V / R so 5 / 8 = 0.625 or 625 mA. For this reason you will either need an amplifier, or a high impedance input. A home stereo RCA input resistance is approximately 1K ohms, which will allow you to connect it ( 5 / 1K = 5 mA ), but the volume level would be too high, and it might be more sensitive to the PWM frequency overtones. You could connect the output through the passive filter on the GinSing circuit into a voltage divider and then into the RCA jack with the least amount of noise. The best solution would be a digital multi-pole filter IC with a cutoff at just under 1/2 the PWM frequency for the best possible sound quality.

The 'I am GinSing' was produced directly from the GinSing shield output using the headphone jack connected via RCA cable to a standard A/D USB converter; so you should easily be able to produce this quality.
The administrator has disabled public write access.
  • Page:
  • 1
Time to create page: 0.185 seconds