The rankings mimics CSRankings to generate an ordered list of researchers in speech/spoken language processing along with their possible research topics, based on publications on important venues of the field in recent five years, so as to help students seeking for PhD studies to find desirable advisors.
The pre-generated report is available at here. To build it by yourself,
- Run
prepare_data.pyto buildpublications.jsonandauthors.json, or simply use the data provided in this repository, covering publications from 2014 to 2024. - Run
export.pyto generate the report.
We scrape author metadata and publication data of the following three types of venues from DBLP, including:
- Speech venues: Interspeech, Speech Communications; SLT, SSW, ASRU, IWSLT (these four are are supported but not included in the pregenerated report)
- Mixed venues: ICASSP, TASLP
- General venues: NeurIPS, ICML, ICLR, ACL(-Findings), EMNLP(-Findings), NAACL(-Findings), KDD, AAAI, IJCAI
All publications in Speech venues are included. Paricularly for Interspeech, section/field of each paper are collected from ISCA Archive to show possible research topics of each researcher. So are the keywords from IEEE Xplore for papers published on IEEE-held venues. Keywords (as well as titles) are also used to filter out non-speech papers in Mixed venues by a set of rules. Titles are used to identify speech papers in General venues. Researchers are sorted by the total number of publications.
The collected data contain errors. The project is by no means an appropriate measure to rank or compare the researchers, and the collected publication data are incomplete as well. Hence the generated index is for reference only and should not be taken seriously.