Take the 2-minute tour ×
Musical Practice & Performance Stack Exchange is a question and answer site for musicians, students, and enthusiasts. It's 100% free, no registration required.
  1. I would like to find out a way to diagnose my vocal problem and assess my voice quality, such as too little or much nasal sound, wrong pitch, weak or loud sound, .... I would like to find out what aspects of my voice need to be improved, before I am able to search for some ways to address them. Now I am not able to find a voice teacher, so I record my voice in Audacity, and hope to compare it to some others' voices which I think are great. But I don't know how to compare mine to others'. Can you provide some tips for me?

    For example, I record myself saying the word "titillate", followed by a very nice recording of the same word which I found from the internet and which I hope my voice can be trained to be as closer to as possible. The recording is here. Following is the waveform (on the top) and spectrogram (on the bottom) of the recording shown by Audacity (for a larger version, click here):

    enter image description here

    My recording is the first half, and someone else's is the second half. Some impressions I have are

    • In our waveforms, mine looks sparser than the other's.
    • in our spectrograms, the white areas are distributed differently.

    What do those indicate? What are your impressions?

  2. Also I would you to help me to identify, solve or improve my vocal problems, after your hear recordings of my speaking. For your better assessment of my voice quality and possible advice to improve it, I record my speaking of a longer passage here.

Thanks!

share|improve this question
2  
I believe the "sparseness" of your recording is due to low audio quality. And in general, I don't think you're going to make much progress with this approach. The timbre of your voice is mostly determined by your physiology and the vast majority of what you see in a waveform, besides frequency and amplitude, is not something you can reliably pick out and work on. You should probably be asking how to solve your vocal problems, rather than deciding on a method and then trying to make it work. –  Matthew Read Dec 30 '13 at 21:22
    
I use Audacity to record my speaking and the playing of someone else's audio. In other words, ours are recorded both by Audacity. So I think sparseness should affect both of us, if it is due to low audio quality? –  Tim Dec 30 '13 at 21:26
    
I edit my post to ask how to solve my vocal problems or improve them. –  Tim Dec 30 '13 at 21:28
    
@Tim - Recording quality is contingent upon the microphone you use, not the program into which the audio files are recorded, though settings in the program can be adjusted. Also, you don't take into account whether or not the other recording has been digitally enhanced / edited to sound more resonant / warmer / etc. –  jjmusicnotes Dec 30 '13 at 21:52

3 Answers 3

A few things:

This is questionably on-topic, because the material in your question seems to address your accent much more than it does the actual technique of vocal production. You didn't record yourself singing, for example, so I am presuming that's not really what you're looking for.

That said, much of the discussion here applies equally well to music (or at least a similar question about music), so I do have an answer for you:

Looking at a spectrogram is not really going to help much for two reasons:

  1. Unless the microphones and recording environments are absolutely identical, there's no way to control your analysis for those variables. Your brain is actually much better at intuitively "upsampling" a recorded voice than "you" will be when looking at a visual representation, no matter how many digital effects you apply.
  2. Using your sense of sight to analyze an aural phenomenon is far less efficient than using your sense of hearing. We're still in the domain of human perception here--using tools like spectrograms is only going to become more efficient when you're dealing with things outside of human perception (like dog whistles or whalesong, for example). Music (or human speech) by definition falls within the domain of human perception (don't tell John Cage I said that), so I can't see a spectrographic analysis being particularly useful here.

Now, my guess is that you're going to the spectrograph because you think your own ears need a little bit of help. This may be the case, but training your ears to be more sensitive will help tremendously both in analyzing and in teaching yourself to improve.

You are on the right track with recording your own voice and comparing it to another speaker. The next step is to start taking notes and experimenting with your own voice to discover what you can change and what differences you can hear. Once you can hear all the differences and you've learned how to change your own voice, then it's just a matter of making the changes you want, and practicing until it becomes habit.


My approach to this would be to listen to a lot of different accents, giving yourself a lot of practice in identifying the specific things that are different in each one. You could also pick up a book like this one (parts of which are viewable here) or this one for a very technical and in-depth overview of the mechanics.

Ultimately, having a teacher is the most efficient way about this, and it must be real-time (in person or over Skype or something) since there is a pretty constant feedback loop that needs to happen, and uploading recording after recording to an asynchronous teacher won't even compare to listening to your own recordings without a teacher--not that that's a bad idea at all!

share|improve this answer
    
Thanks! I agree my ears need to be trained to be able to find out the problems of voices. Do you know if there are some online sources with audios of voices having different problems? –  Tim Dec 30 '13 at 23:14
1  
I wouldn't worry so much about identifying "problems" per se, since a difference is only a problem if you were trying for something else. There are thousands of dialects each with their own distinct characteristics and differences. I believe the books I linked above come with audio recordings, but you can actually find a ton of examples just on YouTube--everything from people blogging from different parts of the world to actors and voice actors doing demo reels of a bunch of different accents at once. –  NReilingh Dec 30 '13 at 23:30

Accent troubles. I can personally relate. I lived in Indonesia for 3 years and I can speak Indonesian fluently; I needed to learn to speak without an accent so I could get parts in movies as a character other than the evil foreigner (that's usually the case).

I've got some suggestions for you and note that I'm just adding on to what NReilingh says because he is also correct.

  1. Immerse yourself in the accent you want to learn. This is extremely important since it transforms your brain faster, but there is a catch; you MUST PAY ATTENTION to the sounds people say when they speak--learning requires focus. This is why many people will live forever with an accent--they don't put effort into learning the proper sounds of the language. I don't blame them for it, but the most common reason is that after they get to a point when their accent is sufficient to be understood, all extrinsic motivation stops; once people can understand you, they will stop correcting you.

  2. Ask people to help you improve your accent when talking to them. Get them to say a word and say it back to them and ask them if it is correct. Yes, this sounds tedious and troublesome. A lot of people are too shy or embarrassed to do this, but I can assure you, the more mistakes you realize you make, the more mistakes you can try to fix.

  3. Once you have a good idea of how the language is supposed to sound, get a good microphone like the Yeti Silver (that's what I have). Get a good pair of over ear headphones. Plug it into the Yeti microphone and you will get live feedback of what you sound like to others. For a while there were certain vowel sounds that I didn't realize I was messing up, but it helped once I heard myself from a recording. Well, instant feedback helps more than that because you can play around with your vocal cords until you find the right sound.

  4. Be patient and keep at it until you get it. I had a lot of motivation to do it and practiced a lot until I could be heard as a native speaker. Sometimes, there are still certain intonation tendencies which I mess up, but that all just comes with more practice.

Good luck. P.S. Check out this forum for people learning English as a second language http://ell.stackexchange.com/questions

share|improve this answer

I think the main difference I heard in the "Titillate" recording was in recording quality, not voice quality. So .. here's a note about the qualities of the "Titillate" recording, and your longer recording :

The second "titillate" sounds like it's quite processed - possibly even from a phone line. The effects I think I can hear are..

  • Some equalisation to pick out the middle range more, whcih generally works well in making speech easily understandable if there is background noise. However it can sound a bit "tinny" like it's being played through a poor quality speaker.
  • Compression - this smooths out the volume so loud parts are quiter, quiet parts louder, meaning if you fumble a syllable it may help pick it up or if one part is louder then it'll smooth it out. This is also used for speech on the radio a lot so that you can easily understand what people are saying in a noisy environment like a car. It can also give a "warm" quality to the voice, making it sound clear and even louder, when it's actually not that loud.
  • Little/no reverb : sounds like either it was recoirded where there is little natural reverb or close up to the microphone (not generally a good idea)

Your own recording has different qualities :

  • No compression (or mild?) meaning it's less "smooth", so doesn't havethat warm quality as mentioned above.
  • Lots of reverb form the room (I assume) meaning it sounds like you're speaking in an echoey room, whcih masks the detail of the voice a bit.
  • Much nicer equalisation (or none applied?) Your recording sounds fuller and more natural :-)

If you wanted to change the way your voice sounds by changing the way you record it, I'd recommend trying this:

  • Record in a room whcih has very little reverb. You can test this by clapping once and listening to how much the sound reverbarates afterwards. Ideally, probably hardly any is best for speech. You can reduce background reverberation by getting closer to the microphone, but I wouldn't recommend that because you start to get popping (peaks in volume) with "P" and "B" sounds. Bear in mind you can add reverb artifiicially if need be.

  • Experiment with addimg compression. I don't know whether Audacity has this (suspect it does) - try adding varying amounts. There's a whole separate discussion on how this works but it'd be hard to go into that here because it depends a lot on the facilities you have available.

  • Equalisation (eq): Sounds like that's quite good already, but there's a mushiness from the reverberation which is at the lower middle range. It might be worth reducing that range a little.

I hope this helps ! I realise you were asking about changing your voice quality, but I wonder whether you really need to do that- just amend the recording method

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.