I don't know anything about that project, but I know a lot about audio electronics...
the wav file is having too much noise
how to record filtered proper voice samples ?
It's best to prevent the noise in the first place. Noise reduction and filtering can sometimes work and it's easier if you have a constant low-level background noise. If the noise is bad, sometimes there's nothing you can do. And you really need a computer, so you'd have to transfer the file to a computer for noise reduction.
Pros still record in soundproof studios with good equipment. On-location movie dialog is re-recorded in the studio. ...Even with pro software there is only so much you can do with a bad recording.
I have some questions, and I'll suggest a couple of troubleshooting experiments you can do -
What's the nature of the noise?
...It's it a low-frequency hum?
...A buzz?
...Hiss?
When do you hear it?
... During silence?
... During quiet parts?
... During loud parts?
The website says:
With only 8-bit sample depth, the signal-to-noise ratio (SNR) is roughly around 40dB (cassette tape), so there’s a bit of background hiss, but if you record audio that has serious dynamic range compression, you don’t notice it (AC/DC’s Back in Black, for example, sounds great).
Back In Black is only going to "sound great" if it's properly normalized. That means the signal level can't be too high (too loud) or the file will be clipped (distorted). If the signal is too low, quantization noise becomes a problem (especially at 8 bits).
If you rip a CD or convert an MP3 to 8-bit WAV, the levels will be correct. But, if you are recording analog it's tougher to get the levels right. And, if you are recording with a microphone, you've got microphone quality, speaker quality, room acoustics, and room noise, all those things degrade sound quality.
Quantization noise is most noticeable during quiet parts. There is no quantization noise with dead-digital silence, and the noise is masked (drowned out) by loud sounds. So, a "constantly loud" song like Back In Black can sound OK, but a classical song with quiet parts and loud parts isn't going to sound good in 8-bits.
i brought a sound sensor for recording voice samples
http://www.amazon.in/LM393-Detection-Sensor-Module-DC4-6V/dp/B00NR3VOJ8?tag=googinhydr18418-21
It says it's a "sound detection sensor". It doesn't actually say it that it puts-out "clean" audio for recording. It also doesn't say if the output is properly biased for the Arduino (which cannot accept the negative half of an AC audio signal).
Do you have a link to the datasheet and/or schematic?
Experiments:
If you can read/write the SDcard on your computer, play the file on your computer. If it sounds bad, you have a recording problem. If it sounds OK, you you have a playback problem.
Also, write a known-good WAV file from the computer to the SDcard to check playback quality. If you don't already have an audio editor [u]Audacity[/u] (FREE) can take an existing WAV or MP3 and convert it to an 8-bit mono WAV.
Connect the line-output or (headphone output) from your computer's soundcard to the input of your Arduino recorder (disconnect the microphone board). Check to see if you can get a clean recording when you bypass the microphone board.
If you have a desktop computer with a regular soundcard, plug the microphone board into the line-input and make a recording on the computer. The mic input on a laptop won't work with the amplified output. Of course, you'll have to power the microphone board. If you can get a clean recording on the computer using the microphone board, the microphone board is OK.