Learn how to fine-tune and serve Faster Whisper Turbo in this comprehensive 35-minute video tutorial. Explore the process of transcribing audio files and YouTube audio using a Colab notebook. Gain insights into the workings of Whisper Turbo and compare it with Faster Whisper and Insanely Fast Whisper. Master the techniques for fine-tuning Whisper Turbo to recognize new words or accents, and discover how to automate training data cleanup using LLMs. Dive into chunking input audio and text data, pushing to hub, and setting up LoRA and Trainer. Finally, learn to save, evaluate, and convert the model for OpenAI format and Faster Whisper, and set up a Faster Whisper Server Endpoint for efficient deployment.
Fine-Tuning and Serving Faster Whisper Turbo
Overview
Syllabus
Whisper Turbo Fine-tuning and Serving
Colab Demo: Transcribing Audio Files and Youtube Audio
How does Whisper Turbo work?
Faster Whisper, Insanely Fast Whisper, and Fast Whisper Server?
Fine-tuning Whisper Turbo for new words or accents
Automating training data cleanup with LLMs
Chunking our input audio and text data and pushing to hub
LoRA and Trainer Setup
Saving, evaluating and converting the model for OpenAI format and Faster Whisper
Setting up a Faster Whisper Server Endpoint
Taught by
Trelis Research
Related Courses
-
Multimodal Audio and Text Fine-tuning with Qwen - Implementation Guide
-
AI Avatars with Dreambooth Alternative LORA Based Stable Diffusion Fine-Tuning on Free Colab
-
Fine-Tuning Llama 3 on a Custom Dataset for RAG Q&A - Training LLM on a Single GPU
-
Fine-Tuning LLMs with PEFT and LoRA
-
Fine-tuning Llama 2 for Tone or Style Using Shakespeare Dataset
-
Full Fine-tuning LLMs with Lower VRAM: Optimizers, GaLore, and Advanced Techniques
