Our team used HearThis for recording Mark. But then after recording they had suggestions for revisions (since the book had been completed 10 years ago). These revisions were made, and included some punctuation changes, which caused problems in the timings. Based on other posts here (such as “HearThis to SAB sound file misalignment”), I was able to use the “Shift Clips” feature, thankfully, to fix the major problems.
But there are still some problems, and I am trying to understand the underlying issues, in order to make the corrections in the right place, so that any future changes in HearThis would not require all of the same corrections. What confuses me is that in HT, everything sounds correct, all recordings are matched with the yellow text lines accurately. But then when using “Export Sound Files” (and choosing SAB, and Audacity, with phrase-level labels selected), it appears that the timing files were not updated to match. Also the *.info text file in the Program Data folder has some problems in the verses that were changed (e.g.the text and line numbers don’t match what shows in HT).
I have then tried to use the “Fine Tune Timings” feature in SAB, but I don’t understand how that works, i.e. which files is it correcting, how those fine tunings are saved, and will those adjustments be saved in case future corrections are made in HT, or will all of those fine tunings have to be done again? And some are not “fine”, since there are still some places where a whole segment/clause is not highlighted in the SAB app.
So I’m wanting to understand the “best practices” in making corrections in the right place after rerecording segments. (It is possible that I messed things up when using “Shift Clips” and renaming a few *.wav files, and now I don’t know how to correct those in the place where it will count.)
Thanks for posting your question. I’m glad you are finding these tool useful and that you were able to use the shift clips feature (somewhat at least) successfully.
Some types of changes to the text (e.g., actual wording changes) will obviously require re-recording clips. Where the changes are merely to the spelling of words, but the recorded words are pronounced correctly, no change is required. (We have a new feature mostly complete, but stuck in a bit of a log jam, that will help you find and deal with those kinds of changes so that the .info files match.) Punctuation changes tend to be the trickiest because as you noted, the recording—at least the combined one for the chapter—can sound fine, but things do not always align. If each clips seems to align to the highlighted text in HearThis, then no shifting or re-recording is necessary. In those cases, the .info file should have the correct information except in places where you manually renamed .wav files. If it does not, please submit a bug report to hearthis_issues@sil.org with details.
Even when everything is perfect in HearThis, there is still the possibility of the highlighting being off in SAB. One cause of this that is easy to fix is if the phrase breaking settings in SAB are not set to match the phrase breaking settings in HearThis. When you export to SAB in HearThis, I think the output window shows a reminder indicating what phrase breaking characters to use in SAB. So that is the first thing to verify.
Even with that correct, there are still some subtle differences between HearThis and SAB that require HearThis to guess a little bit at the timings. Most of time, these result in relatively small misalignments, but sometimes they can be noticeable. Unfortunately, if you use “Fine Tune Timings” in SAB, there is no way for the adjusted timings to find their way back into HearThis, so if you later make additional changes in HearThis, you’ll have to start over in SAB. (I’m not sure how easy this would be to do in SAB but you can maybe just reimport only the changed chapters, if you know which ones they are, so you would only have to redo those chapters.) I am not aware of any reason why entire segments should be getting skipped in SAB if everything is aligned correctly in HearThis, so if that is happening, again please report it as a bug and we will request the necessary information to follow up.
A final option for getting things to align in SAB is to use Aeneas. In my experience, it is not quite as good as the alignments produced by HearThis when everything is working correctly, but for most languages it is amazingly accurate, so it might be an adequate solution. If you try it and are dissatisfied, you can always pull the HearThis timing files back in and go back to figuring out the problems.
If you need additional help, please reply and we will try to find a time where we can work together to sort things out.
Thanks for your thorough suggestions. I was in a bit over my head, and it took a long time to get back at this, but now most things are worked out, with the yellow highlighting showing nicely in the app. Some corrections were done with the Fine Tune Timings feature, and some (where there were audio issues) in Audacity.
One place we still have an issue that I can’t figure out how to correct, is in a verse where there is a quote from the Old Testament. I suppose that the quote codes are causing issues. There is an intro phrase with just a comma, and then two sentences within the quote. However, the second sentence does not show up in the Fine Tuning chart, and never gets highlighted. That probably would have been ok, but then I tried to be sneaky and change the html file info for that segment, thinking I could trick it into highlight the right phrases for the right timings. But I guess I made it worse. And now I am confused as to where I should really be making the changes, i.e. where that info is actually being stored. (BTW, we had exported the “fine tune timings” files to be worked on in another computer, which added another layer of learning curve and more files to keep track of while experimenting.)
It’s just one verse, so we could just ignore it, since this is really just a trial copy of the app, for a limited number of people. But I would love to learn more about this, in order to make our corrections easier in the future.
(Regarding Aeneas, I am still a little doubtful of how to do it with Thai script. And since we already had it “mostly” marked correctly, we stuck with making adjustments for now.)
I’m not clear as to whether you are talking about an info.html file in HearThis or some HTML file in SAB. Which HTML file did you edit (and where)? Do the two sentences show up to be recorded inside HearThis? If so do they both have recordings that correctly correspond to the text? “Sneaky” editing of files that are under program control and are not intended to be edited by human beings can definitely lead to problems. It’s often not obvious how things are hooked up or how information stored in one file might relate to information stored elsewhere. If you have a copy of the original version and the edited version, please send them to tom_bogle@sil.org and I will see if I can sort things out. Also, if you can define what you mean by “made things worse” that might be helpful.
I am not an Aeneas expert, but I suspect someone has done the work of working out how to make it behave as expected for Thai script. If at some point you want to explore that further, possibly @ChrisHubbard can point you in the right direction.
Thanks. Your response made me aware that my questions are now more related to SAB, rather than the initial HearThis issues. The HTML file referred to is in relation to the timings files in SAB. (We had to make some edits in the recordings after they were exported from HearThis, so we can’t depend on the original HT files any more.) So I will pursue solutions by email or in the SAB posts.