# Low Error Rate

QLoRA fine-tuned version based on LLaMA3 Youko, optimized for Japanese visual novel English translation with 8B parameters

Reverb Speaker Diarization V2 is a speaker diarization model based on pyannote-audio, outperforming the baseline pyannote3.0 model on multiple test sets.

A fine-tuned printed license plate OCR model based on microsoft/trocr-base-printed, with a character error rate of 0.037 on the evaluation set

An Arabic precise diacritization system based on byte-level fine-tuned models, automatically completing Arabic text diacritics by fine-tuning pre-trained models.

A deepfake audio detection model fine-tuned based on Microsoft's WavLM-base, achieving excellent performance on the ASVspoof 2019 dataset with an accuracy of 99.79%

An audio classification model based on the HuBERT architecture, specifically designed for detecting deepfake audio and audio spoofing

A Chinese speech recognition model fine-tuned and optimized based on whisper-large-v3, showing significant performance improvements in multiple Chinese speech benchmarks

Transformer-based OCR model for Spanish printed text, optimized for printed fonts and does not support handwriting recognition

An OCR model fine-tuned based on microsoft/trocr-base-printed, specifically designed for recognizing printed license plate numbers.

An OCR model fine-tuned based on microsoft/trocr-base-printed, specifically designed for recognizing license plates and timestamp information

A speaker embedding model based on ResNet293 architecture, optimized with large margin fine-tuning, supporting tasks such as speaker recognition, similarity calculation, and speech segmentation

A fine-tuned German speech recognition model based on Whisper Large v3, optimized for German speech processing and recognition

A captcha recognition model fine-tuned based on Microsoft's trocr-base-printed model, specifically designed for OCR tasks involving printed text

This model is fine-tuned on the Common Voice, JVS, and JSUT datasets for Japanese speech recognition tasks using openai/whisper-base.

This model can convert images of handwritten mathematical expressions into corresponding LaTeX sequences, suitable for mathematical formula recognition and digital processing.