Towards Multi-Level Transcript Segmentation: LoRA Fine-Tuning for Table-of-Contents Generation
By: Steffen Freisinger , Philipp Seeberger , Thomas Ranzenberger and more
Potential Business Impact:
Organizes spoken words into chapters for easier understanding.
Segmenting speech transcripts into thematic sections benefits both downstream processing and users who depend on written text for accessibility. We introduce a novel approach to hierarchical topic segmentation in transcripts, generating multi-level tables of contents that capture both topic and subtopic boundaries. We compare zero-shot prompting and LoRA fine-tuning on large language models, while also exploring the integration of high-level speech pause features. Evaluations on English meeting recordings and multilingual lecture transcripts (Portuguese, German) show significant improvements over established topic segmentation baselines. Additionally, we adapt a common evaluation measure for multi-level segmentation, taking into account all hierarchical levels within one metric.
Similar Papers
Text to Trust: Evaluating Fine-Tuning and LoRA Trade-offs in Language Models for Unfair Terms of Service Detection
Computation and Language
Helps computers find unfair contract rules faster.
MeTA-LoRA: Data-Efficient Multi-Task Fine-Tuning for Large Language Models
Computation and Language
Teaches AI more with less information.
Robust Federated Finetuning of LLMs via Alternating Optimization of LoRA
Machine Learning (CS)
Makes AI learn faster and better with less data.