Generating Structured Summaries from Video Transcriptions

Generating Video Summary

Welcome back! In the previous lessons, we explored transcribing videos using the gpt-4o-transcribe API and downloading them via both Google Drive and LinkedIn. Building on those skills, we're now going to delve deeper into generating video summaries, an essential skill for transforming lengthy transcriptions into concise and insightful content. This lesson takes you a step further by utilizing the capabilities of OpenAI's API to create detailed yet succinct summaries on the fly.

What You'll Learn

Today, you will: Understand how to generate a summary from a transcription. Explore the use of the OpenAI's API to craft well-structured summaries. Gain insights into designing effective prompts for better results. Learn about system and user roles in OpenAI API requests and their importance.

Understanding Generating Video Summary

Summarizing transcriptions involves distilling the core messages from extensive spoken content, ensuring key points are retained while unnecessary details are filtered out. When dealing with long videos or lectures, extracting the main themes allows you to quickly grasp the essentials without listening to every word. The OpenAI API facilitates this by leveraging advanced language models capable of understanding context and summarizing long texts. In our context, the OpenAI API will help us convert raw transcriptions into coherent summaries. Here's an overview of a prompt used within the API to achieve this: Python{ "role": "system", "content": ( "You are an expert content analyst and summarizer with these capabilities:\n" "- Extracting key points while maintaining context\n" "- Identifying main themes and core messages\n" "- Preserving critical details while reducing length\n" "- Maintaining the original tone and intent\n" "- Organizing information hierarchically\n\n" "Format your summaries with:\n" "1. A one-sentence overview\n" "2. 2-3 key takeaways\n" "3. Important details or quotes (if any)" ) }{ "role": "system", "content": ( "You are an expert content analyst and summarizer with these capabilities:\n" "- Extracting key points while maintaining context\n" "- Identifying main themes and core messages\n" "- Preserving critical details while reducing length\n" "- Maintaining the original tone and intent\n" "- Organizing information hierarchically\n\n" "Format your summaries with:\n" "1. A one-sentence overview\n" "2. 2-3 key takeaways\n" "3. Important details or quotes (if any)" ) } The system prompt sets the stage by providing essential guidelines to the model about its role and the expected output format. This enables it to focus and execute tasks effectively. Let's also examine the user prompt, which specifies the task at hand: Python{ "role": "user", "content": ( f"Create a structured summary of this transcription. " f"Focus on the core message and key points while maintaining " f"context and critical details.\n\n" f"Transcription:\n{text}" ) }{ "role": "user", "content": ( f"Create a structured summary of this transcription. " f"Focus on the core message and key points while maintaining " f"context and critical details.\n\n" f"Transcription:\n{text}" ) } The user prompt defines the specific requirements, such as the transcription text that needs summarization, ensuring the model understands and processes the content correctly.

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

Examples and Step-by-Step Explanations

Let's walk through summarizing a transcription using OpenAI's API. First, let's see what a typical transcription and its summary look like in practice.

Here's an example of a video transcription and the structured summary it produces:

Original Transcription:

Welcome everyone to today's presentation on sustainable energy solutions. My name is Dr. Sarah Chen and I'll be discussing three major renewable energy technologies that are transforming our energy landscape. First, let's talk about solar power. Solar panels have become incredibly efficient over the past decade, with modern photovoltaic cells converting up to 22% of sunlight into electricity. The cost has dropped by over 80% since 2010, making solar the cheapest form of electricity in many regions. Second, wind energy continues to grow rapidly. Modern wind turbines can generate up to 15 megawatts of power, and offshore wind farms are particularly promising because ocean winds are more consistent and stronger. Finally, battery storage technology is solving the intermittency problem. Lithium-ion batteries have improved dramatically, and new technologies like solid-state batteries promise even better performance. The key takeaway is that renewable energy is not just environmentally necessary, but economically viable. We're at a tipping point where clean energy is becoming the dominant choice for new power generation worldwide.

Welcome everyone to today's presentation on sustainable energy solutions. My name is Dr. Sarah Chen and I'll be discussing three major renewable energy technologies that are transforming our energy landscape. First, let's talk about solar power. Solar panels have become incredibly efficient over the past decade, with modern photovoltaic cells converting up to 22% of sunlight into electricity. The cost has dropped by over 80% since 2010, making solar the cheapest form of electricity in many regions. Second, wind energy continues to grow rapidly. Modern wind turbines can generate up to 15 megawatts of power, and offshore wind farms are particularly promising because ocean winds are more consistent and stronger. Finally, battery storage technology is solving the intermittency problem. Lithium-ion batteries have improved dramatically, and new technologies like solid-state batteries promise even better performance. The key takeaway is that renewable energy is not just environmentally necessary, but economically viable. We're at a tipping point where clean energy is becoming the dominant choice for new power generation worldwide.

Generated Summary:

Dr. Sarah Chen presents an overview of three transformative renewable energy technologies that have reached economic viability and are reshaping the global energy landscape.

Key Takeaways:
1. Solar power has achieved remarkable efficiency gains (up to 22% conversion) and cost reductions (80% decrease since 2010), making it the cheapest electricity source in many regions
2. Wind energy, particularly offshore wind farms, offers consistent power generation with modern turbines producing up to 15 megawatts
3. Battery storage advances, especially in lithium-ion and emerging solid-state technologies, are solving renewable energy's intermittency challenges

Important Details:
- "Renewable energy is not just environmentally necessary, but economically viable"
- We've reached "a tipping point where clean energy is becoming the dominant choice for new power generation worldwide"

Dr. Sarah Chen presents an overview of three transformative renewable energy technologies that have reached economic viability and are reshaping the global energy landscape.

Key Takeaways:
1. Solar power has achieved remarkable efficiency gains (up to 22% conversion) and cost reductions (80% decrease since 2010), making it the cheapest electricity source in many regions
2. Wind energy, particularly offshore wind farms, offers consistent power generation with modern turbines producing up to 15 megawatts
3. Battery storage advances, especially in lithium-ion and emerging solid-state technologies, are solving renewable energy's intermittency challenges

Important Details:
- "Renewable energy is not just environmentally necessary, but economically viable"
- We've reached "a tipping point where clean energy is becoming the dominant choice for new power generation worldwide"

As you can see, the summary maintains the core message while condensing a 200+ word transcription into a structured format that highlights the essential information.

Implementation Process

Building our code on top of prior lessons' knowledge and extending it to a new dimension of handling video content will lead us to success. Here's part of the implementation process:

Python

from openai import OpenAI

class MediaProcessorService:
    def __init__(self):
        self.client = OpenAI()

    def summarize_transcription(self, text):
        try:
            response = self.client.chat.completions.create(
                model="gpt-4o",
                messages=[
                    {
                        "role": "system",
                        "content": (
                            "You are an expert content analyst and summarizer with these capabilities:\n"
                            "- Extracting key points while maintaining context\n"
                            "- Identifying main themes and core messages\n"
                            "- Preserving critical details while reducing length\n"
                            "- Maintaining the original tone and intent\n"
                            "- Organizing information hierarchically\n\n"
                            "Format your summaries with:\n"
                            "1. A one-sentence overview\n"
                            "2. 2-3 key takeaways\n"
                            "3. Important details or quotes (if any)"
                        )
                    },
                    {
                        "role": "user",
                        "content": (
                            f"Create a structured summary of this transcription. "
                            f"Focus on the core message and key points while maintaining "
                            f"context and critical details.\n\n"
                            f"Transcription:\n{text}"
                        )
                    }
                ]
            )
            return response.choices[0].message.content
        except Exception as e:
            print(f"Error generating summary: {e}")
            return None

from openai import OpenAI

class MediaProcessorService:
    def __init__(self):
        self.client = OpenAI()

    def summarize_transcription(self, text):
        try:
            response = self.client.chat.completions.create(
                model="gpt-4o",
                messages=[
                    {
                        "role": "system",
                        "content": (
                            "You are an expert content analyst and summarizer with these capabilities:\n"
                            "- Extracting key points while maintaining context\n"
                            "- Identifying main themes and core messages\n"
                            "- Preserving critical details while reducing length\n"
                            "- Maintaining the original tone and intent\n"
                            "- Organizing information hierarchically\n\n"
                            "Format your summaries with:\n"
                            "1. A one-sentence overview\n"
                            "2. 2-3 key takeaways\n"
                            "3. Important details or quotes (if any)"
                        )
                    },
                    {
                        "role": "user",
                        "content": (
                            f"Create a structured summary of this transcription. "
                            f"Focus on the core message and key points while maintaining "
                            f"context and critical details.\n\n"
                            f"Transcription:\n{text}"
                        )
                    }
                ]
            )
            return response.choices[0].message.content
        except Exception as e:
            print(f"Error generating summary: {e}")
            return None

Here's how the process flows:

Initialization: The OpenAI API is initialized with an implicit API key provided in the OPENAI_API_KEY environment variable, allowing access to its functionalities.
Transcription generation: Using the gpt-4o-transcribe API we covered in previous lessons, we generate a text transcription for the given video.
Summarization: The summarize_transcription method leverages an OpenAI LLM, GPT-4o, to process the text. It sets system and user prompts to guide the summarization, maintaining context and detail precision.
Output: The returned summary is a concise version of the transcript, highlighting the content's main points.