LibriCSS recipe #4321
Merged
LibriCSS recipe #4321
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
kkm000
requested changes
Nov 11, 2020
egs/libri_css/s5_mono/local/convert_rttm_to_utt2spk_and_segments.py
Outdated
Show resolved
Hide resolved
|
@kkm000 Thanks this was very helpful! I have made (almost) all the changes you suggested. |
|
Thanks! @kkm you can merge if you think it's OK.
…On Thu, Nov 12, 2020 at 5:03 AM Desh Raj ***@***.***> wrote:
@kkm000 <https://github.com/kkm000> Thanks this was very helpful! I have
made (almost) all the changes you suggested.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#4321 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAZFLO2KUUGPLEFILTL5UCLSPL32ZANCNFSM4TKU5VHQ>
.
|
kkm000
requested changes
Nov 12, 2020
egs/libri_css/s5_mono/local/diarization/post_process_css_rttm.py
Outdated
Show resolved
Hide resolved
You make me 😳 :) By the way, there is a nice book called Pure Bash Bible. A lot of good tricks! |
kkm000
requested changes
Nov 12, 2020
|
I'm fine with hosting it but please prepare the doc files that are needed.
y.
…On Thu, Nov 12, 2020 at 10:57 PM Daniel Povey ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In egs/libri_css/s5_mono/local/data_prep_css.sh
<#4321 (comment)>:
> + wget -c --load-cookies /tmp/cookies.txt \
+ "https://docs.google.com/uc?export=download&confirm=$(wget --quiet \
+ --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate \
+ 'https://docs.google.com/uc?export=download&id=1Piioxd5G_85K9Bhcr8ebdhXx0CnaHy7l' \
+ -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1\n/p')&id=1Piioxd5G_85K9Bhcr8ebdhXx0CnaHy7l" \
+ -O for_release.zip && rm -rf /tmp/cookies.txt
I think it's a good idea to mirror on OpenSRL but we should check what
@jtrmal <https://github.com/jtrmal> thinks.
In any case, IMO we shouldn't let this delay merging.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#4321 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACUKYXY4SCFZLM7JHCOG5SLSPSVCZANCNFSM4TKU5VHQ>
.
|
kkm000
approved these changes
Nov 13, 2020
|
Thank you @desh2608, that was a helluva work! LGTM, merging. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This recipe contains multi-speaker diarization + ASR for long recording. There are 2 scenarios:
s5_mono: The input is single-channel mixed recording (possibly containing overlapping speech).s5_css: The input is continuously separated audio streams. By "continuous" separation, we mean a method that performs separation on windowed segments, such that the same speaker may be split across different audio streams in different windows.For diarization, we use x-vectors and spectral clustering. For ASR, we use a TDNNF-based model (pretrained model from Librispeech) fine-tuned on Librispeech + simulated RIRs. Evaluation is in terms of cpWER (similar to CHiME-6). More details are in the README.md file.