Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

grammar #15

Open
wants to merge 4 commits into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
31 changes: 31 additions & 0 deletions src/content/en/docs/AI Tools/whisper/_index.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,31 @@
---
title: Whisper AI
weight: 4
categories: [Tools]
tags: [WhisperAI]
description: >
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification. You can generate transcripts from audio files.
---

|Responsible AI Checklist|
|---|---|
|Can be used with customer data| ❌ No|
|Can the output be used at customer| ✅ Yes|
|Can be used with Xebia Internal secret data| ❌ No|
|Can the output be used commercially by Xebia| ❓TODO VERIFY PROOF |
|Is data being stored in the region you use it | ❌ No|
|Do we have a Xebia license / managed solution for this tool | ❌ No|

## What is Whisper AI
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification. You can generate transcripts from audio files.

## How to use Whisper AI
You can visit [the website][1] to follow instructions. In short. You download the tools and run whisper on your commandline. You can use multiple language models. The bigger the model, the more accurate the transcript. Although in many cases, small or medium is sufficient.

## License / Costs
whisper is free

## Suitable to use with clients?
It generates a transcript from an audio file. It does so locally. But please check customer. Also make sure the transcript does not end up somewhere.

[1]: https://github.com/openai/whisper
Original file line number Diff line number Diff line change
@@ -0,0 +1,72 @@
---
title: Creating an Annotated speaker deck for training purposes from scratch
weight: 4
categories: [Presentation]
tags: [PowerPoint, SlackGPT, WhisperAI]
description: >
How can you create a PowerPoint deck with speaker notes so your presentation or training can be redelivered?
---

|Responsible AI Checklist|
|---|---|
|Can be used with customer data input| ❌ No |
|Can the output be used at customer | ❓ TODO check this|
|Can be used with Xebia Internal secret data| ✅ yes|
|Can the output be used commercially by Xebia | ✅ yes |
|Is data being stored in the region you use it | ❌ No|
|Do we have a Xebia license / managed solution for this tool | ✅ [SlackGPT](/docs/ai-tools/slackgpt)|
|Tools used in workflow | [SlackGPT](/docs/ai-tools/slackgpt), [Whisper AI](/docs/ai-tools/whisper) |

## Introduction
When you create a PowerPoint deck that can be used in a training or other sort of multi-delivery, it is hard to capture the story you want to convey in slides only. What do ou tell at every slide, what are the talking points and what story line do you want to follow. When you deliver the presentation yourself, this is doable, but what if someone else needs to use your deck? Or you deliver a training deck to a customer. For example, a customer wants a deck they can use for training their people. Then you want to annotate the slides with speaker notes, the story you want to tell so that people can read it, and the talking points so people can create their own story without losing the storyline.

No better way to capture this by recording the presentation right? Right! Bu that does not really scale well when preparing or delivering. So you want your story in the speaker notes.. But that is a lot of work! Yes indeed. And that is where this workflow comes in

## Step 1: Deliver your presentation to your recording device
Deliver the presentation. On Windows, use the audio recorder and deliver your story as you would do it in a training or presentation. Start every new slide with <Slide 1>, <Slide 2>

## Step 2: Create a transcript with Whisper AI
The audio file, wav, m4a etc has all the content. Install WhisperAI and run the tool by using the commandline

```
whisper youraudio.m4a --model medium
```

You can mention teh language, although whisper can detect this as well. It can take some time to do this. As a workaround, create a VM in the cloud and let it run there.


## Step 3: Style it

With the transcript that is generated by WhisperAI you can go to SlackGPT and use the following prompt per slide (manual copy paste)

```
I'll provide you with a transcript. You will start with a header "Talking Points" and then need to summarize the transcript into bullet points and a short descriptive text of the point underneath. The description of talking points need to be in blog style. Described as an outside observer. Write clear and use imperative tone. Only use information that is provided in the transcript.
Then, add a header "speaker text" and write a text that someone can read to present this to an audience. Do not use announcer speech or things as today. Treat this as it is in the middle of a presentation
Here is the transcript:
```

### Examples

```
I'll provide you with a transcript. You will start with a header "Talking Points" and then need to summarize the transcript into bullet points and a short descriptive text of the point underneath. The description of talking points need to be in blog style. Described as an outside observer. Write clear and use imperative tone. Only use information that is provided in the transcript.
Then, add a header "speaker text" and write a text that someone can read to present this to an audience. Do not use announcer speech or things as today. Treat this as it is in the middle of a presentation
Here is the transcript:

So let's get started with an introduction to
GitHub Advanced Security for Azure DevOps or GHAZDO.
So in this module, we will talk a little bit about the context of why we
need things like advanced security and why this
is becoming much more important than it maybe was in the past.
```

Output
```
Talking Points
- Introduction to GitHub Advanced Security for Azure DevOps (GHAZDO)
* In this part of the presentation, the speaker introduces the audience to GitHub Advanced Security for Azure DevOps, also known as GHAZDO. This topic is crucial to understand due to the increasing importance of advanced security in today's world.
- The context of advanced security
* The speaker will discuss the reasons behind the growing need for advanced security measures, and why it is more important now than ever before.
Speaker Text
Now, let's dive into the topic of GitHub Advanced Security for Azure DevOps, or GHAZDO. In this module, we'll explore the context behind the increasing demand for advanced security and why it has become a critical aspect of our digital lives.
As technology continues to evolve, so do the potential threats and vulnerabilities we face. This makes it essential for us to understand and implement advanced security measures to protect our systems and data. So, let's discuss the importance of advanced security and how GHAZDO can help us ensure a safer digital environment.
```