Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[SOLVED] Frame division in a speech signal

Status
Not open for further replies.

Shweta_S

Member level 3
Joined
Jun 20, 2006
Messages
59
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,288
Location
Chennai
Activity points
1,655
Why do we have to divide the speech signals into frames of 10ms or 20 or 25ms? How did the value come to the picture? Please help me... !!!
 

tedy58

Newbie level 2
Joined
Jan 10, 2012
Messages
2
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,283
Activity points
1,291
The explanation is simple,

there is direct connection between time and frequency.
for example if
tf=0.020; i.e. time frame is 20 ms then symbol-rate is 1/tf
1/20 ms = >> 1/tf

ans =

50
then You can count the number of elements in the frame, i.e nf=fix(tf/ts); i.e.
>> nf

nf =

144
or in a frame of 20ms You will have 144 discrete elements of the signal.
 
Last edited:

Shweta_S

Member level 3
Joined
Jun 20, 2006
Messages
59
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,288
Location
Chennai
Activity points
1,655
I get what you are saying, we get 144 samples of the signals. My question was why 20ms and not 1ms or maybe 50ms or any other number for that case.
Does that time frame have a particular significance?
 

tedy58

Newbie level 2
Joined
Jan 10, 2012
Messages
2
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,283
Activity points
1,291
When You take frame of 1 ms your frame will have only 14,4 samples. And for analyzes purposes', it is needed not less than 90 (or 180 ) samples in advance at LPC. this is why we take 20ms, 22,5ms or higher values.
 

Shweta_S

Member level 3
Joined
Jun 20, 2006
Messages
59
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,288
Location
Chennai
Activity points
1,655
Thank you.

I got some information too regarding this question!

The fundamental frequency of speech usually ranges around 100Hz.
So we take 20ms frames so that in this time frame only one phoneme occurs and atleast 2 periods of the phoneme occur in this time frame.

Basically our vocal chord can't vibrate faster than 10ms. So we resort to 20ms time frame.
 

er.deepshikha

Newbie level 6
Joined
Dec 2, 2011
Messages
12
Helped
5
Reputation
6
Reaction score
2
Trophy points
1,283
Activity points
1,349
The explanation is simple,

there is direct connection between time and frequency.
for example if
tf=0.020; i.e. time frame is 20 ms then symbol-rate is 1/tf
1/20 ms = >> 1/tf

ans =

50
then You can count the number of elements in the frame, i.e nf=fix(tf/ts); i.e.
>> nf

nf =

144
or in a frame of 20ms You will have 144 discrete elements of the signal.

hi, please tell me what is ts
thnx in adv
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top