AIbase
EN
Home
Xlsr Sg Lm
X

Xlsr Sg Lm

Developed by manifoldix
Swiss German speech recognition model fine-tuned on FHNW v1 Swiss Parliament dataset
Speech Recognition
Transformers
 Other#Swiss German Speech Recognition#Dialect Robustness#Parliament Scenario Optimization
Downloads 17
Release Time : 3/2/2022

Model Overview

This model is an automatic speech recognition (ASR) model optimized for Swiss German, particularly suitable for Swiss parliamentary scenarios and dialect recognition. Based on the XLS-R-1b architecture, fine-tuned on 70 hours of Swiss parliamentary data.

Model Features

Swiss Dialect Support
Specially optimized for Swiss German dialects, achieving a 40% word error rate on dialect test sets
Parliament Scenario Optimization
Fine-tuned using 70 hours of Swiss parliamentary data, achieving a 34.6% word error rate on parliamentary test sets
Large Model Architecture
Based on the XLS-R-1b architecture with powerful speech feature extraction capabilities

Model Capabilities

Swiss German Speech Recognition
Dialect Speech Recognition
Parliament Scenario Transcription

Use Cases

Government Agencies
Parliament Meeting Minutes
Automatically transcribe Swiss parliamentary meeting content
Word error rate 34.6%
Dialect Research
Swiss Dialect Analysis
Identify and transcribe various Swiss dialects
Word error rate 40%
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase