Thread: Speech Analysis

fread() reads in size * NumberOf bytes:

fread(bufferArrayName, Size, NumberOf, FilePointer);

smart to take it one value at a time, I suspect.

Is this file a WAV file with raw PCM data?

I found this tidbit interesting:

BhupathiRao

hai everyone; I am reading 'time001.wav' sound file by using wavread function. But it gives the amplitude values between [-1 to +1]. How can i read amplitude values more than the [-1 to +1].
More options Apr 24, 5:03 am
Newsgroups: comp.soft-sys.matlab
From: "BhupathiRao " <[email protected]>
Date: Sat, 24 Apr 2010 09:03:04 +0000 (UTC)
Local: Sat, Apr 24 2010 5:03 am

Subject: how to read amplitude out of range in my wav file

hai everyone;

I am reading 'time001.wav' sound file by using wavread function. But it gives the amplitude values between [-1 to +1]. How can i read amplitude values more than the [-1 to +1].

Rune Allnor

More options Apr 24, 6:46 am
Newsgroups: comp.soft-sys.matlab
From: Rune Allnor <[email protected]>
Date: Sat, 24 Apr 2010 03:46:57 -0700 (PDT)
Local: Sat, Apr 24 2010 6:46 am

Subject: Re: how to read amplitude out of range in my wav file

On 24 apr, 11:03, "BhupathiRao " <[email protected]> wrote:

> hai everyone;

> I am reading 'time001.wav' sound file by using wavread function. But it gives the amplitude values between [-1 to +1]. How can i read amplitude values more than the [-1 to +1].

As I understand it, the contents of the .wav file is normalized
to the [-1,1] interval. If correct, it means that you can not read
values outside that interval.

Rune

Wayne King
More options Apr 24, 6:58 am
Newsgroups: comp.soft-sys.matlab
From: "Wayne King" <[email protected]>
Date: Sat, 24 Apr 2010 10:58:06 +0000 (UTC)
Local: Sat, Apr 24 2010 6:58 am

Subject: Re: how to read amplitude out of range in my wav file

"BhupathiRao " <[email protected]> wrote in message <[email protected]>...
> hai everyone;

> I am reading 'time001.wav' sound file by using wavread function. But it gives the amplitude values between [-1 to +1]. How can i read amplitude values more than the [-1 to +1].

Hi, if the data are represented by 32 bits or less in the wave file and you read the file with the default 'double' format, the data range is always [-1,1]. If you read the data as 'native' you will get a range depending on the number of bits.

At any rate, I think you will have to scale the result in Matlab to get a signal with values outside of [-1,1].

If you created the file by sampling audio and you know the voltage input range at the A/D converter, say [-5,5] volts. Then you can read the file using 'native' and use the quantization step to convert the values to volts in the Matlab command window.

Wayne

Forward

So are we looking for a -1, 1 range or -5, 5, or ??

Okey, this reads in one sample at a time, but, I would choose to use short. I stop by three times before I reach the sound buffer and get info from the header at offsets found on the internet.

Code:

#include <stdio.h>

#define FILENAME "omega.wav"
#define SRATE_OFFSET 24
#define BITS_OFFSET 34
#define SIZE_OFFSET 40


int main(void)
{
        int x;
        short clip_info[10000];
        int sampleRate, bitDepth, dataSize;
        FILE * file_in;


        file_in = fopen(FILENAME,"r");

        if (file_in == NULL)
        {
                printf("Error Opening file.");
        }
        else
        {
                printf("success\n");
                fseek(file_in, SRATE_OFFSET, SEEK_SET);
                fread(&sampleRate, 4, 1, file_in);

                fseek(file_in, BITS_OFFSET, SEEK_SET);
                fread(&bitDepth, 2, 1, file_in);

                fseek(file_in, SIZE_OFFSET, SEEK_SET);
                fread(&dataSize, 4, 1, file_in);

                fread(clip_info, 10000, 1, file_in);
        }

                for (x=0; x<= 5000; x++)
                        printf("% d\n", clip_info[x]);

                printf("Sample rate: %d\n", sampleRate);
                printf("Bit depth: %d\n", bitDepth);
                printf("Sound data size: %d\n", dataSize);

        return 0;
}

BTW, no error checking and so going on here.

Yeah, that doesn't look right. It looks right here, but I use a different sound file. Did you get rid of the loop in fread? I made some edits to the above code to reflect what I did here.

Other issues could be that there is no error checking, for the reads which should be in there to confirm that the data is read successfully before printing etc. Or, that you are using a non standard wav but a broadcast wav, they have a large meta data chunk in the header. Since you recorded this yourself you can confirm that with the the program you did the recording in.

Later on you would probably also want to use malloc to allocate memory for the sound.


Edit: BTW, do you print to screen or do you write it to your text file only? The problem could be when you are writing it back to the text file as well, if you are not redirecting the output to a file in a terminal.