# Automatic Speech Recognition / Machine Translation systems by UPV

- Institution: CERN
- Department: IT Technical Forum (ITTF)
- Taught by: Jorge Iranzo, Joan Albert Silvestre, Jaume Santamaria
- Year: 2025
- Language: English
- Source: [CERN](https://videos.cern.ch/record/3021361)
- Licence: [CERN](http://copyright.web.cern.ch)
- Who can watch: Teens profiles and older (sources give no ratings; placed by Invisible Network from the source's own words)

The Universitat Politècnica de València (UPV) as a CERN contractor has provisioned an on-premises Automated Speech Recognition (ASR) and Machine Translation (MT) solution from February 2022 to August 2024.

More precisely, the Machine Learning and Language Processing (MLLP) research group of the UPV, with proven expertise in the development of language processing technologies for high quality media transcription and translation, developed English and French ASR and MT systems adapted to the particle-physics domain for offline and streaming multimedia content.

MLLP's transcription and translation services and documentation were provided to the CERN via an on-premises installation of the MLLP's Transcription and Translation Platform (TTP). MLLP's ASR and MT services once adapted to the particle-physics domain thanks to the tight collaboration with CERN Media Services proved to supersede the performance of publicly available large pretrained ASR and MT models.

Finally, CERN was provided with an auto-training procedure to automatically enhance ASR systems on a regular basis by taking advantage of newly produced resources and materials at CERN. This talk describes the development, evaluation and deployment of these particle-physics adapted ASR and MT systems now available to the CERN community..

0:00:00 Slide 1

0:03:23 Slide 2

0:05:15 Slide 3

0:06:31 Slide 4

0:13:37 Slide 5

0:16:06 Slide 6

0:18:05 Slide 7

0:19:40 Slide 8

0:21:29 Slide 9

0:23:18 Slide 10

0:25:10 Slide 11

0:27:01 Slide 12

0:29:47 Slide 13

0:35:34 Slide 14

0:37:34 Slide 15

0:42:25 Slide 16

0:44:46 Slide 17

0:48:10 Slide 18

0:50:07 Slide 19

0:53:05 Slide 20

0:55:16 Slide 21

0:57:35 Slide 22

1:06:52 Slide 23

1:08:41 Slide 24

1:10:44 Slide 25

1:14:25 Slide 26

1:16:18 Slide 27

1:21:18 Slide 28

1:30:20 Slide 29

1:32:18 Slide 30

## Lectures

1. [Automatic Speech Recognition / Machine Translation systems by UPV](https://invisiblenetwork.ai/learn/cern/3021361/e0350ccc9a50), 1 h 34 min (CERN's player)