SCROLL

Voice Editing

Use Sound Forge as an example to demonstrate how to edit the sound sources.

STEP 1
Open the original sound file as below
STEP 2
Convert the sound file to the format that we can accept (WAV and MONO)
STEP 3
Normally, we will have high frequency noise introduced when we resample the original sound file. This high frequency noise can be reduced by using the Equalizer function as shown below:

EQ setting example:

60 HZ or below Depend on situation, sometimes noise will be in this frequency rang, therefore, we need to decrease the gain of this frequency range
150 HZ Dog bark, Drum
400 HZ Male speech
1000 HZ High pitch male, female speech
2400 HZ High pitch female
6000 HZ Children
15K HZ or above Usually, we will reduce the gain of this frequency range into minimum because our voice chip cannot response to this.

The value setting for the above commands are not fixed. User should choose the suitable value for his/her particular sound file.

STEP 4
The sample rate is dependent on which IC body is used. We have the following IC bodies:
OTP Chip Memory Size Calculation of Sampling Rate
Sample Rate = Memory Size / (voice duration x Data encoding algorithm )
10 sec 256K ex sound total 5sec , S.R. = 256K / (5sec x 4bit) ≒ 12.8KHz ( 4bit ADPCM )
ex sound total 5sec , S.R. = 256K / (5sec x 5bit) ≒ 10.2KHz ( 5bit u-Law )
ex sound total 5sec , S.R. = 256K / (5sec x 8bit) ≒ 6.4KHz ( 8bit PCM )
21 sec 512K ex sound total 8sec S.R. = 512K / (8sec x 4bit) ≒ 16KHz ( 4bit ADPCM )
ex sound total 8sec S.R. = 512K / (8sec x 5bit) ≒ 12.8KHz ( 5bit u-Law )
ex sound total 8sec S.R. = 512K / (8sec x 8bit) ≒ 8KHz ( 8bit PCM )
42 sec 1M ex sound total 12.5sec S.R. = 1M / (12.5sec x 4bit) ≒ 20KHz ( 4bit ADPCM )
ex sound total 12.5sec S.R. = 1M / (12.5sec x 5bit) ≒ 16KHz ( 5bit u-Law )
ex sound total 12.5sec S.R. = 1M / (12.5sec x 8bit) ≒ 10KHz ( 8bit PCM )
85 sec 2M ex sound total 28sec S.R. = 2M / (28sec x 4bit) ≒ 18KHz ( 4bit ADPCM )
ex sound total 28sec S.R. = 2M / (28sec x 5bit) ≒ 14.2KHz ( 5bit u-Law )
ex sound total 28sec S.R. = 2M / (28sec x 8bit) ≒ 8.9KHz ( 8bit PCM )
170 sec 4M ex sound total 70sec S.R. = 4M / (70sec x 4bit) ≒ 14.2KHz ( 4bit ADPCM )
ex sound total 70sec S.R. = 4M / (70sec x 8bit) ≒ 7.1KHz ( 8bit PCM )
341 sec 8M ex sound total 153sec S.R. = 8M / (153sec x 4bit) ≒ 13KHz ( 4bit ADPCM )
ex sound total 153sec S.R. = 8M / (153sec x 5bit) ≒ 10.4KHz ( 5bit u-Law )
ex sound total 153sec S.R. = 8M / (153sec x 8bit) ≒ 6.5KHz ( 8bit PCM )
11 min 32M ex sound total 235sec S.R. = 32M / (235sec x 8bit) ≒ 17KHz ( 8bit PCM )

After obtaining the sample rate, we should resample the original sound. If Sound Forge is used, and assume the new sample rate is 12KHz, then, we can resample the sound file using ‘Effect’ and ‘Resample’ to modify the sample rate as shown in the figures bellows:

STEP 5
Convert the sound file to the format that we can accept ( 8-bit )
STEP 6
Remove the silence portion at the begin and end of the sound can reduce the chance of “POP” sound to happen.Finally, If there are very low volume part in the middle of the sound, it should be converted to total silence otherwise, the low volume sound maybe become noise sound in the final IC.
STEP 7
save as
KEEP LIVING FANTASTIC

Copyright © APLUS INTEGRATED CIRCUITS INC.