New transcribed dataset: MLS

nmstoker · December 16, 2020, 1:02am

This new transcribed dataset looks like it could be very helpful - I don’t know how feasible (or desirable) it would be to incorporate it into training for the next release of the main English model, but with 44.5k hrs transcribed it compares well to the amount of transcribed audio on the earlier LibriSpeech dataset (1,000 hrs)

It also has quantities of transcribed audio for other languages too but those are less dramatic (but could still be a big help compared with what’s available for them too)

synesthesiam · December 16, 2020, 3:39am

Awesome, thank you for the link!

eziolotta · January 18, 2021, 10:12am

Hi, We have implemented an MLS Importer for the Italian speech dataset.

github.com/MozillaItalia/DeepSpeech-Italian-Model

MITADS-Speech/mls_importer.py

master


#!/usr/bin/env python3

import time
import os
import re
from corpora_importer import ArchiveImporter,Corpus

CORPUS_NAME = 'mls'

#function fix_apostrophe and load_mailabs_fixed_token
##see this: https://github.com/MozillaItalia/DeepSpeech-Italian-Model/issues/124
def fix_apostrophe(text_normalized,fixed_token):
    
    tokens = text_normalized.split()
    for tok in tokens:

        if(tok in fixed_token):
            ##replace works only if there are no ambiguous tokens like : destro ->d'estro  sera->s'era
            text_normalized = text_normalized.replace(tok,fixed_token[tok])

This file has been truncated. show original

The work is not yet finalized and we would like to try various training tests.
MLS has audio clips ranging from 10 to 20 seconds.

github.com/MozillaItalia/DeepSpeech-Italian-Model

LIST OF AUDIO+TEXT DATASETS

opened 06:09PM - 17 Dec 20 UTC

nefastosaturo

help wanted dataset

## LIST OF ALL ITALIAN DATASETS FOUND From issue #90 I'm putting here all the d…atasets that have been discovered. Some of them are plug-and-play for Deepspeech others instead need to be created from scratch (splits up audio by sentences) Feel free to pickup one that has not been done for checking it out. ### NOTE If one of this dataset needs a deeper analysis please do not start a discussion here but open a new issue and I will update this table with the issue reference. ## DATASETS | dataset | hrs | url | plug-n-play | TODOs | doing | done | note | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | ------------- | |**MLS** | 279.43 h | [↗](http://openslr.org/94/)| | | | | **HOT!!!!** |VoxForge #111 | 20h | [↗](http://www.repository.voxforge1.org/downloads/it/Trunk/Audio/Main/16kHz_16bit/)| ✔ | <ul><li>- [x] url replace in DS import_voxforge.py script</li><li>- [x] fix import sys error </li></ul> | ✔ | | |MAILABS | 127h40m | [↗](https://www.caito.de/2019/01/the-m-ailabs-speech-dataset/)| ✔ | | | ✔ | |Evalita2009 | 5h | [↗](http://www.evalita.it/2009/tasks/digits)| | | | ✔ | |MSPKA | 3h | [↗](http://www.mspkacorpus.it/)| | | | ✔ | |SIWIS | 4.5h | [↗](https://phonogenres.unige.ch/index.php?page=téléchargement)| | | | ✔ | |SUGAR | 1.5h | [↗](https://github.com/evalitaunina/SUGAR_Corpus)| | | | | sentences are not useful |VociParlateWikipedia #34 | ? | [↗](https://it.wikipedia.org/wiki/Categoria:Voci_parlate)| |<ul><li>- [ ] sync audio with its page revision</li></ul> | | | |EMOVO | ~12m | [↗](http://voice.fub.it/activities/corpora/emovo/index.html)| |<ul><li>- [ ] align filename codes with their sentences </li></ul> | | | interesting for emotions (disgust, happy..) |ZIta | <1hr | [↗](https://github.com/ChMeluzzi/ZIta)| | | | | transcriptions do not follow recordings (eg: Lett_Z_Sp1_zero.wav) |LIM_Veneti | <1hr | [↗](https://github.com/ChMeluzzi/LIM_Veneti)| | | | | no audio files? |split-MDb | ~46m | [↗](http://www.parlaritaliano.it/index.php/en/corpora/644-spit-mdb-spoken-italian-multilevel-database)| |<ul><li>- [ ] parse&clean the .wrd files </li></ul> | | | based on CLIPS |tg60 | 1h30m | [↗](http://www.parlaritaliano.it/index.php/it/dati/650-corpus-di-parlato-telegiornalistico-anni-sessanta-vs-2005)| |<ul><li>- [ ] long audio files to be split </li></ul> | | | maybe among the info files there are some timings that could be useful for splitting up? |PraTiD | 1h12m | [↗](http://www.parlaritaliano.it/index.php/en/corpora/645-corpus-pratid)| |<ul><li>- [ ] long audio files to be split </li></ul> | | | From CLIPS; maybe among the info files there are some timings that could be useful for splitting up? |ParlatoCinematografico | ? | [↗](http://www.parlaritaliano.it/index.php/it/dati/659-corpus-di-parlato-cinematografico)| |<ul><li>- [ ] long audio files to be split </li></ul> | | | .lab files with speakers timings |PerugiaCorpusPEC | ? | [↗](https://www.unistrapg.it/cqpwebnew/)| | | | | a login is needed. License?

lissyx · January 18, 2021, 11:40am

Please share that as a PR as soon as you can!

eziolotta · January 18, 2021, 2:45pm

MLS, and all importers we are doing recently, use our utility for common operations (corpora_importer.py), they depend on this utility.
Then we have a collector for generating a final speech dataset that aggregates all imported corpora.
DeepSpeech EN repo use a different strategy.
Should I also do a PR of the utilities?

In any case, the work on our corpora_collector has yet to be completed, hopefully soon.

lissyx · January 18, 2021, 2:50pm

That’s a good question, maybe it would be useful to move to that factorized code as well, what do you think @reuben ?