Crowdsourcing a Multi-lingual Speech Corpus: Recording, Transcription and Annotation of the CrowdIS Corpora