For those of you who want to actually measure this according to common practice, here's a link to a phonetics course as USC that uses Praat.

http://sail.usc.edu/~lgoldste/General_Phonetics/Week10/Formant_Analysis/index.html

That's a great link! Thanks for posting it!

Yeah the formants do only seem to behave properly with ideal condidions. There is so much variance otherwise.

Super hard to just test for resonance given a regular clip of talking.

It takes a bit of practice to find the ideal points for the measurement but everything goes pretty smoothly once you get the hang of it. I'll post the bits that I can help with in the video... as soon as I get to it.

hmm - I tried recording some vocals and analyze them - I got pitch about same, but when looking at the f3, I get 2700 for a, but 2900 for e and 31oo for i. So is that regular that vocals have different f3?
Those articles are very nerdy, I dont think I got them - But I guess they support the thing about looking at the f3 and how to measure it for resonance. I think it did not give ranges for M and F, but Jenny posted them before from memory - like <2600 for M and >3000 for F with a gender neutral part in the middle (where I obviously am - again)

Quote from: anjaq on October 09, 2013, 01:42:39 AM
hmm - I tried recording some vocals and analyze them - I got pitch about same, but when looking at the f3, I get 2700 for a, but 2900 for e and 31oo for i. So is that regular that vocals have different f3?

Quote from: Jennygirl on October 05, 2013, 03:45:27 AM

In this chart, male F3 falls in between ~2,250hz and ~2,900hz depending on the vowel. Female F3 on the other hand falls anywhere from ~2,600hz to ~3450hz. My speech therapist told me something similar to that with males at 2k-2.5k and females from 3k-3.5k.

Each individual chart corresponds to a different vowel sound. So if you test "i", the first chart could give you something to test against. Because yes, F3 varies depending on the vowel sound. Think of how your mouth/throat changes to form sounds.. sometimes it's getting smaller in certain areas and opening up in others, and that's how you form a vowel.

If you say "i" as in "ee", your tongue comes up in the back of your mouth which reduces the amount of airspace...  That gives it a slightly higher resonance than other vowels by default.

Seems better to test specific vowels instead of just getting the average F3 for an entire passage.

And I'm definitely curious to see some more statistics on female F3's. I will try to hunt them down, but if anyone has clues it is all appreciated!

This is a great tutorial! Can it be stickied? Or maybe somehow incorporated into the wiki? How does that work anyway?

Not sure if this can be considered necroposting if it's a sticky, but this caught my interest so I decided to download the program and give it a shot.

With absolutely no warm up, I captured a sample of middle C with the "ahh" vowel sound.  I was trying to be on the quiet side and I didn't really warm up first so it sounded awful, but it worked well enough.  The pictured sample gave an average f3 of ~3262.35.  If I am browsing the above correctly, this is good?

Hi Delsorou!

No, you don't have to worry about necropost on this one.

Anyway, your question makes me wonder. I kinda want to go back for a refresher from that voice therapist! But yes based on the rudimentary explanation she gave me, that would be correct I think. 3262hz seems good to me.

Maybe someone who has more experience can chime in. I still think the more subjective approach is best when it comes to resonance.. I.e. you could post a sample in the does my voice pass thread and get some feedback.

Best wishes to you

Thanks for the tutorials Jenny

The formant portion is really helpful for narrowing down what works and what doesn't once you listen to yourself and figure out the sound you hit.

The formant/resonance is odd though as you mentioned before. I tend to hit the same resonance on the vowels when I tested.

Awesome! Thank you so much n__n

My normal speaking voice (I did the Rainbow Passage) was at 75Hz. I did it again and tried to raise my voice a little and it showed 100Hz.

Now I'm sad. 

I wonder how accurate that Formant reading is. We did a bit of praat analysis on each others and our own voice recordings in the susans chatroom on the weekend and there were some voices that had a very high number of T3 , like 3200, but they did have a sound that felt like it was not quite there yet, while others had 2600 or 2800 and sounded really good... So maybe it is a bit of a rough measurement and if it is below 2500 it really is not that good but above that it depends? On what? . hmmm


Something else I noticed - how to work around that, would be a question. If I read or speak something and analyze it in praat for average pitch, it highly depends on me speaking either more monotonous or with more intonation and melody to the voice. Just speaking more melodious will increase the average pitch by 30 or 50 Hz at times. I thus try to find parts of the recording where I say "errrr" or "hmmmm" or end words with a vowel and try to find the more common pitches or the lower pitches to get an idea of the "baseline". Does that make sense? Otherwise I guess just adding melody to my voice would bring it from 120-130 Hz of my more relaxed monotonous voice into 180Hz range which is in the female range - but it still has these low drops which make people notice something is at odds.
So which parts of the recording is good to analyze - the overall average or only some parts of it?

EDIT: Here a praat image of what i mean. Same text, same base pitch level and level or relaxation but first part of the recording was not emphasizing on voice melody and the second part was really using voice melody as it probably should be. It made a 50 Hz difference!




Thanks

PRAAT is good for measuring an input file, but I would recommend GRAM if you need a real time visual aid for pitch etc, during practice.
GB

Yes indeed, Praat is more a very powerful tool for analyzing recordings while GRAM (Or rather "Spectro-GRAM", I think) is great for real time analysis. It is similar to "Overtone Analyzer".

Does the 'Pitch Range' setting in Pitch Settings help for ignoring the weird recording artefacts?  Even with my studio mic, every recording has the lines that go all the way to the top and bottom when looking at the pitch.

If you're out and about and need to check your pitch before an interview, date. Try " DaTuner light"  on your mobile. It doesn't record but it does give real time analysis.