search

RoboTutorLLC / RoboTutor_2019

Main code for RoboTutor. Uploaded 11/20/2018 to XPRIZE from RoboTutorLLC/RoboTutor.
Other
7 stars 4 forks source link

1.8.9.1 Eliminate static from songs #324

Open sarahfialko opened 6 years ago

sarahfialko commented 6 years ago

Rerecord 9 songs with static from 3/19/18 https://drive.google.com/drive/u/0/folders/1F_4t0RnztmiEW98hxlUfRiOlcNdFJAoJ

JackMostow commented 6 years ago

Is Filipo available? If not, can Judith sing them?

sarahfialko commented 6 years ago

Filipo is away for the summer. @judithodili: Jack asks if you can sing them.

judithodili commented 6 years ago

Naaaaa.... We can wait for Filipo... lol... or Leonora can sing them too... but I think we can wait for Filipo/

On Thu, May 31, 2018 at 12:37 PM, Sarah Fialko notifications@github.com wrote:

Filipo is away for the summer. @judithodili https://github.com/judithodili: Jack asks if you can sing them.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/RoboTutorLLC/RoboTutor/issues/324#issuecomment-393594878, or mute the thread https://github.com/notifications/unsubscribe-auth/ACiS-Bk7QJzSfe6v54qu5X3uDFJt1OYKks5t4BxMgaJpZM4UOVgh .

--

Regards, Judith Odili Uchidiuno www.judithu.com

JackMostow commented 6 years ago

@nirmalpatel - Do you have quick and easy audio magic to find and mitigate the annoying bursts of static in the 12 songs at SWAHILI EDUCATIONAL SONGS? If not, it should take only a few hours to re-record them.

nirmalpatel commented 6 years ago

@JackMostow man, this was something. I got :zap: :zap: so many times due to those statics and my audiophile headphones.

I have reuploaded the main folder with all the fixes here: https://drive.google.com/drive/u/1/folders/1hz_hAJ3xbyx4TGCOLduD6FWJBcTwjrnc

I ended up fixing 32 audio files. I went through each folder, listened to each audio file, and removed the static from every file that had it. 2 of the 32 files had people talking in them, so I fixed that too. I have fixed the files to a reasonable quality, and I think they all can be used in the RT now.

This CSV file contains the list of files I have fixed: https://drive.google.com/open?id=1Voye_FwZqHB4tGqSQSq_bbpySjTzJmXa

image

In each folder within the main folder, there is a cleaned folder which contains fixed audio files. Audacity projects related to each of the files are besides the actual audio files. I have left the buggy files as they are. For example, consider this folder:

RT11-POSEIDON A E I O U 111351.576 03192018/

Here is the cleaned folder:

image

Within the cleaned folder there are fixed WAV files: image

And Audacity projects related to those WAV files are also there, in case someone wants to re-work on them:

image

Please let me know if you have any questions.

-- Nirmal

JackMostow commented 6 years ago

@nirmalpatel - THANK YOU! Your fix sounds neither quick nor easy but is much appreciated.

@kevindeland -

  1. (Where) are the audio files for the Swahili songs in GitHub other than in RoboTutorLLC/CodeDrop1_Assets?

  2. Should we leave CodeDrop1_Assets unchanged and put new and changed content elsewhere? If so, where, and who should copy it and put the updated audio files there?

@judithodili - The fixed files are on Google Drive in SWAHILI EDUCATIONAL SONGS AUDIO STATIC REMOVED with their original Swahili titles prefixed by RT11-POSEIDON, the machine on which they were narrated, e.g. RT11-POSEIDON ba be bi bo bu 123738.214 03292018.

However, they have different folder names in CodeDrop1_Assets/CodeDrop1_Songs/assets/audio/sw/cmu/xprize/story_reading/quality_low/songs/, e.g. LC_Vowel_Song_1.

  1. Do you know the mapping from the folder name of each song in GDrive to its folder name in GitHub?

  2. Who should copy the fixed folders to the right place in GitHub?

Thanks. - Jack

judithodili commented 6 years ago

Thanks for removing the static Nirmal .... while this removes the static, it may have created another issue...

The json file for each story uses the start and stop times for each word during recording to decide when to highlight a word (based on the .seg file) ... even making small edits to the original audio files will throw this highlighting off.

Nevertheless, I think we should try the songs using these new audio files to see if it terribly throws off the highlighting... if not, problem fixed! Otherwise, we might need to re-record..

Thank you Nirmal ... Jack, please can you also assign this to Kevin? I’m replying from my email so can’t tag. --

Regards, Judith Odili Uchidiuno www.judithu.com

JackMostow commented 6 years ago

@judithodili - @kevindeland is already an assignee.

@nirmalpatel - Did you leave the audio durations unchanged, or did you splice out the static segments, in which case how long were they? The difference would cause the highlighting to lag by their duration, but if that's the most broken thing about RoboTutor, we're in terrific shape!

nirmalpatel commented 6 years ago

@JackMostow yikes! So I have spliced out the statics, but they are very small segments, so I am guessing that it won't be a really major issue. For some files, where the beginning of the files was the static part, I might have spliced off a bit too much because there was a bunch of space around the static to select and delete. But I don't think there are more than 5 files where I have done so.

So it would be great if someone could test out the files that I have exported.

JackMostow commented 6 years ago

Testing the fixed audio in RoboTutor needs to wait until:

  1. @kevindeland decides which repo to put the songs in, and
  2. Someone (probably @judithodili or I) figures out which audio folder in GitHub to update for each song fixed in GDrive, because their folder names differ.
JackMostow commented 6 years ago

@kevindeland - where should I put the fixed audio files after figuring out which folder is which?