Feature Suggestion: Accent Training with Audacity

Special Interest Groups Audio Processing
1

So I bought American Accent Training. It seems to me that the biggest challenge in learning any accent is comparing one’s own voice with a native speaker. There is no software that allows you to do that in a time efficient manner.

I am looking into an easy way to select an audio segment from a long audio track. I get those tracks from YouTube after I download the video using Download Helper extension for Firefox then I get them imported via FFmpeg tools into Audacity. As I play the file I can easily zoom-in to an audio segment and select it. Audacity is really great with all of this!

Now I need to record my voice. Play the original track then play the recorded track in an automated fashion. This gives me the chance to compare my own voice with the native speaker. Then I redo the recording process till I get my voice to sound as close as possible to the speaker.

The problem right now with Audacity is I have to listen to a segment, record my voice, mute the original, listen to my recording, mute mine, listen to original, then delete the track I just recorded then re-record again.That’s too cumbersome to do!

Is it possible to implement a language training mode or something that does the following?

  1. Listen to a selected audio segment
  2. Record to a separate track using the same length as the selected segment or a bit longer
  3. Play original segment only (automatically mute all other channels)
  4. Play recorded track only (automatically mute original channel)
  5. Re-record on the recorded track (not the original one)
  6. Go to 3 (Loop)

Right now, I select a segment, export it to separate audio file then load it in AnkiSRS memorization software. It takes too long to do that and I could really spend the time learning American accent rather than creating the files. This new mode helps a lot with accent training for any language. I teach at the university and having a good accent is easier for my students to understand me and also helps when I give presentations.

Can you help please?

Thank you.

2

Can you make any use of Multitrack Overdubbing? After you get your system set up, you speak at the exact same time as the trainer playback and your voice goes down on a separate rack. From then, you can fade back and forth between the two and listen, syllable by syllable to the differences. That’s something that can be done now with existing software and maybe a little more hardware.

With the extra hardware, you can hear yourself in real time in your headphones versus the trainer as you’re making the dual track recording.

Koz

3

The original purpose of this technique is one person in the basement (optional) recording a symphony by playing each instrument one after the other and overlaying them.

4

Any update on this? I agree with kozikowski about the use of multi-track for recording different instruments.
But for learning an accent you need to keep listening a particular paragraph or a sentence then record your own then play the original then the recorded. Muting and unmuting and recording a new track every time is very time consuming and a deterrent. Thanks

5

Maybe I have not explained it clearly. Here is what I’m trying to do exactly.
“I tried a simple sound recorder on a computer as well as a tape recorder, and it did not work - too complicated, invoves too much button pressing. You need a program which plays a phrase, then records you repeating it, then repeats the native speaker, then plays your recording, and does it all over again in a loop until you are satisfied. It turned out that it is not easy to find a program like this”

I tried the program in the question but it is not free and does not have any keyboard shortcuts so I ended up clicking so many times but it is the only thing available.
http://www.amazon.com/review/RQEQXMPZUCXQG/ref=cm_cr_pr_viewpnt#RQEQXMPZUCXQG

Any progress on such a feature? Thanks!

6

How much are money are you offering that everything else should be dropped to do this?


Gale

7

Moved to the correct “Adding Features” board.

Posting in my replies already made to the user on feedback@:

  1. Listen to a selected audio segment
  2. Record to a separate track using the same length as the selected segment or a bit longer

There is a menu item Transport > Overdub (on by default). If that item is checked (on), recording will only be made according to the length of any selection made in the track. However the original audio will be heard while you record, which may not be what you want.

With overdub off, you will always have to press Stop to stop the recording manually.

  1. Play original segment only (automatically mute all other channels)
  2. Play recorded track only (automatically mute original channel)

We can add your vote for a button to cycle through the tracks playing each in turn on its own.

If you are not already doing so, press Mute on the original track and leave Mute on assuming you don’t want to hear the original while recording). Then you can hear your
rendition as soon as you have recorded it by pressing Space. Press Solo on the original, then Space to hear that original only, and press Solo again on the original to release Solo
then Space to hear your rendition only.

Alternatively, enable “Simple” mode for the Solo button:
http://manual.audacityteam.org/o/man/tracks_preferences.html#behaviors

then press Solo on each track to solo that track and mute the other track.

  1. Re-record on the recorded track (not the original one)
  2. Go to 3

It would be another (popular) feature request For Audacity to be able to “punch-in” record over existing audio (for example, redo your first attempt at rendition without opening a new track).

But you can Append Record to the end of an existing track:
http://manual.audacityteam.org/o/man/transport_menu.html#append

for your rendition of a new phrase in the track you are already recording.

A way to make this entire process somewhat automated. >Select a segment
manually> Press record then it records voice (add 1 second or half a second
time at both ends of the recording segment, wait a bit, play the original
in solo, play recording in solo, re-record again over the original
recording. This automation would be a huge time saving for such a task.

Yes, but in such a scenario, Audacity cannot predict if you may want to hear the recording in solo two or three times before recording again. It would probably be better to have a shortcut that played each track soloed in turn until you pressed the shortcut again.

Punch-in recording is a separate request. Obviously it is a convenience for a highly repetitive workflow like yours. The arguments against it have tended to focus on the fact that you can generally get a better result in high quality music work by having the corrected recording on its own track and cross-fading it into the original.


Gale

8

In the absence of any further input from the original poster I have archived this thread on Audio Processing.

WC