K

Kotoba Whisper V2.2

Developed by kotoba-tech
Japanese automatic speech recognition model based on Whisper, integrating speaker separation and punctuation addition functions
Downloads 22.80k
Release Time : 10/18/2024

Model Overview

Kotoba-Whisper-v2.2 is a Japanese automatic speech recognition (ASR) model developed based on the Whisper architecture, with added post-processing capabilities for speaker separation and punctuation insertion.

Model Features

Speaker diarization
Incorporates diarizers technology to identify and separate speech content from different speakers
Automatic punctuation
Uses punctuators technology to automatically add punctuation to transcribed text
Efficient inference
Supports Flash Attention 2 acceleration to improve GPU inference efficiency

Model Capabilities

Japanese speech recognition
Multi-speaker separation
Automatic punctuation insertion
Long audio processing

Use Cases

Meeting minutes
Multi-speaker meeting transcription
Automatically identifies speech content from different speakers in meetings and generates punctuated text records
Can distinguish between different speakers and generate formatted meeting minutes
Interview records
Interview transcription
Converts interview recordings into text, automatically distinguishing between interviewer and interviewee speech
Generates interview records with speaker identification and punctuation
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase