Susan's Place Logo

News:

Visit our Discord server  and Wiki

Main Menu

What numbers do I aim for in voice analysis?

Started by Mara, September 10, 2010, 08:41:51 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Mara

Okay, so I got several cool programs that let me talk to my computer and make squiggly lines on a screen!  Yay!  Now what?

Meaning, I'm practicing my voice.  What attributes am I looking at, and what's the female vs. male range for each?  Cause right now, all I have are squiggly lines and some numbers that are meaningless to me.

EDIT: To be clear, I have a couple of spectrogram programs, some of which have something to do with voice.  But I'll probably keep looking for more until I find good ones.  But without knowing what needs to be measured, I can't really tell what kind of program I need.
  •  

pebbles

I've only used spectogram myself. I change the display type to Scope 1 personally so it looks like a line graph.

Depected in this image is a capture of the voice of a Natal lady called Lowery Turner.

The horizontal line indicate pitch in KHz the Vertical line indicates amplitude. The red lines are between 170hz and 230hz they indicate a "catchment target" for me when I'm training Female voice (Although it is entirely possible and maybe desirable to develop a female voice that targets below the basket) and a female vocal inflection requires alot of change in pitch so above and below that is fine however when singing a flat note that is generally where you should be aiming in terms of sheer pitch increase. And I should also point out alot of those really high pitch noises above the 1khz arn't really all that audible to human hearing so really you ought to work most on the voice and structure composition between the 0 and the 1 on the horizontal axis.

however there are significant structural differences between the male and female voices besides from absolute pitch. You can have women with deep female voices.

I made this quite afew months ago my voice has improved since then this was originally for my personal records. So ignore the blue line for now.
The green line in the back of this sample indicates my male voice and was always read as male As you can see compared to the female sample provided by a natal lady The male voice possesses "lower peak" beneath the red line in addition there are many MANY more peaks and thus individual tones in a male voice than a female voice.

If you count them from the 0 to 1 khz range
Male voice = 9
Female voice = 5

(the very low peak right at 0khz sample is actually a static noise from my microphone it's always present)

If you were wondering what the blue line is that was my early attempts at a female voice it was Very high pitch and only possessed 3 tones thus sounded rather shrill. Not male but just a nasty unfufilling noise.

now it looks more like this. Sample taken today

And is structurally much similar to a natal female.

In order to emulate a female voice you need to force close part of your voice box so you artificially shorten your vocal cords.
You have got the muscles to perform this ability but it's just a matter of you learning howto do it.
Some vocal coaches recommend gargling others merely describe it as "talking from behind your male voice"  I don't understand what any of that means that myself
What worked for me is I sung notes with my male voice along the scale of a piano and watched what happened in the window.

After a certain point you will notice the harmonic of your voice change the tension in your voice will disable the lower vocal registers and you will sing the high notes with your upper registers thus eliminating about half of the harmonics from your voice. Then it's a matter of feeling that tension holding it then singing the same effect down the piano keys until it becomes too stressful and your voice hurts what your trying to do is pull this effect down into a normal speaking range ie 170-230 sector after you have that you will kind of sound female however you will still have to work on speech mannerisms and inflection to spit polish up the voice aswell as increasing your stamina and comfort with it

:) I hope this helped I'm not a speech therapist and I've never been officially counciled by one but my voice appears to pass now I've not been He'd on the telephone any time recently and when I speak it seems to help my passablity rather than hinder it.
  •  

azSam

Wow Pebbles. I'm bookmarking that and I'll certainly give it a good read. Thanks for the info!
  •  

Mara

Sorry for the late reply, but I wanted to thank pebbles for posting that. :)  It's VERY helpful.  I wouldn't have known that I could tell by the number of peaks how my voice sounded if not for this.  And the tips on modifying the voice sound good.

I appreciate it!

EDIT: Also, I wanted to give you an applaud but I can't seem to find the arrows anywhere.  Do I have to be at 50 posts or something to give karma?
  •  

Renate

It might be a bit more complicated than the number of peaks in a range.

Pitch is the overall framework within which your voice signal demonstrates periodicity.
The rule of thumb is that 160 Hz is the dividing line between male and female pitch.
Of course you can be a Marlboro-smoking, Jack Daniels-drinking woman
with a pitch a bit lower than Johnny Cash and still be consistently identified as a woman.
Still, for most of us, making an effort to raise the pitch is useful.

The frequencies above the basic pitch are also critical to identification.
These are a bit more difficult to quantify.
Your voice canal is a filter and you want to resonate on the higher frequencies and not on the lower ones.

Yes, this can be as difficult and complicated as it sounds.

For basic pitch analysis see Praat in Susan's Wiki.
  •  

Rosa

Thanks for that link. I've been looking for a free voice analysis program.  Looks a bit complicated though.  Can anyone give basic instructions on how to get the recorded voice to show up on a graph that indicates pitch?  I'm getting graphs, but nothing I can understand.
  •  

Renate

In Praat, go to Read/Read from file.
It will read a WAV file and display the name in the left hand object pane.
Select the object, then click, "Edit" in the right hand verb pane.
A graph will appear.
On the top is a basic waveform.
On the bottom should be just a single blue line indicating pitch.
If there is spectrogram stuff, just go to the menu and unclick Spectrum/Show spectogram
Now you can click anywhere on the bottom section and a vertical cursor will appear and the frequency will read out on the right.

Try recording a sweep, that is, go "oooh" from your lowest note to your highest note.
  •  

Rosa

Thank you Renate!  This works when I record a sentence or a swoop, but when I do a few sentences together I don't see the pitch line, and sometimes on one sentence the line is broken.  Do I need to select a smaller section or something?  Is there a way to see pitch while I am reading so I can practice keeping it up above 160?
  •  

Renate

The line is broken because a lot of what you say is "unvoiced", without really a pitch.
If you click and drag, then you can select a section to average.

There are various spectrum / spectrogram tools for real time.
I don't know any particularly to recommend.
I use a home-brew FFT.
  •