VocaGPT FAQ

Supported Devices and Systems

The App runs on iOS 15+ (iPhone and iPad) and MacCatalyst 15.0+ (MacOS Monterey+).

The App Trial

The App will initially run in trial mode. The user is given an allowance of 10K tokens and a trial period of two weeks. The trial ends when the token budget is exceeded or the two week period is elapsed, whatever comes first.
At the start of the trial period, you will be asked to agree to the App product Terms of Service. You cannot use the App in trial or purchase modes without first agreeing to the Terms Of Service.
You can optionally immediately purchase the App by pressing on the "Buy now / Restore" button.
The provided trial AI Service is Azure OpenAI running the GPT-3.5-turbo model.
You are encouraged to test your own API Keys in VocaGPT during the trial period.

App Trial Completion and App Purchase

Once the trial has ended, if you wish to continue using the App, you need to obtain your own API Key from a supported AI Provider, then purchase the App and finally enter your API Key into App Settings -> Services

Benefits Of The App

The pay-for-use API pricing model is extremely more cost effective than fixed subscription fees.
The convenience of having information available at your fingertips, especially when you are on the go.
Vocalization is a more natural way to express your queries.

Agents

The currently supported Agent roles are: Query (for general queries), Translate: (performs language to language translation and nothing more) and App Support (answers App support queries).
All agents have selectable input and output languages and selectable voices.

AI Services

The currently supported AI Services are: OpenAI(GPT-4, GPT-4o, GPT-4o-mini, GPT-4-turbo, GPT-3.5-turbo), Azure(supporting OpenAI(GPT-4, GPT-4o, GPT-4o-mini, GPT-4-turbo, GPT-3.5-turbo), Google(Gemini 1.5, Gemini 1.0 and PaLM), Anthropic(Claude 3.5 Sonnet, Claude 3.0 Opus, Sonnet and Haiku, Claude 2.1) and MistralAI(Nemo, Large, Open-8x22B, Open-8x7B and Open-7B) More services may be supported in time.
Higher tier models such as OpenAI GPT4 have throttling restrictions that you may experience with repeated use.

Obtaining Your Own API keys

OpenAI

To get your OpenAI API Key, and optional Organization you can follow these general steps:
1. Go to openai.com/api and click on "Get Started" to create an OpenAI account if you don't already have one.
2. Log in to your OpenAI account.
3. On the left hand icon panel mouse-over the Pad Lock icon and the label "API Keys" will appear. Click on the pad Lock icon.
4. On the API Keys page, click the "Create new secret key" button.
5. Enter a name for your secret key and click "Create secret key". Make sure to save this key somewhere secure as it allows access to your OpenAI account.
6. Your new secret API key will now be listed on the API Keys page. Copy this string of characters - This is your API Key that you will enter into VocaGPT in the Apps OpenAI AI Service.
7. If you have an assigned Organization with your account, then that needs to be added to VocaGPT as well. If you do not have an assigned Organization, then leave it blank in VocaGPT.
Azure (OpenAI model)

To get the following: your API Key, Azure Resource, Azure Deployment, The Azure Rest API Version for Azure's OpenAI Models, you can follow these general steps:
1. Go to the Azure portal portal.azure.com and sign in with your Microsoft account credentials.
2. Click on "Create a resource" and search for "OpenAI". Select the "OpenAI" resource from the results.
3. Click "Create" to begin setting up OpenAI on Azure. Choose your subscription, resource group, region, and pricing tier. Click "Review + create".
4. Review the configuration and click "Create". Wait for the deployment to complete.
5. Once deployed, go to the OpenAI resource. So, now you have your Azure Resource name and Azure Deployment name.
6. Click on "Keys and Endpoint" in the left sidebar of the Resource page and copy the API Keys given there.
7. Currently, the preferred Azure Rest API Version is "2023-07-01-preview". You can obtain a list of Azures Rest API versions by searching for "Azure REST API versions (Azure AI Search)".
Google (Gemini)

To get your Google Gemini API Key you have two options:

The simpler option:
1. Go to the Google Makersuite: makersuite.google.com and generate your own API Key.
The formal option:
1. Go to the Google Cloud Console at console.cloud.google.com
2. Create a new project or select an existing project where you want to enable the Gemini API.
3. On the sidebar menu, navigate to "APIs & Services" > "Library".
4. View all the Services for the "Machine learning" category.
5. Select Gemini for Google Cloud.
6. Once enabled, click on "Credentials" in the sidebar menu.
7. Click "+ Create Credentials" and select "API key".
8. Copy the generated API key. This is the key you will add in VocaGPT for a Google Gemini AI Service.
Anthropic (Claude2)

To get your Anthropic Claude, API Key, you need to apply to Anthropic first, you can follow these general steps:
1. Visit Anthropic's web console at console.anthropic.com. The console provides an opportunity to evaluate Claude's capabilities before starting technical integration
2. Apply for Access: Follow the instructions for getting access to Claude, which you can find in the API's documentation or via the web console.
3. When you receive notice of access, go back to the console and generate your API Key.

How to add your AI Service Provider API Key into the App

A quick note about copy and pasting API Keys and information into the App. You can use Apple's Universal Clipboard to copy API Keys displayed on one computer/device and paste onto another device running the App. This really makes the process of entering keys into the App much simpler. See: Unversal Clipboard Support

First, access App Settings by clicking the Gear-Cog icon at the top right of the App's title bar.
In App Settings, scroll down to the Services section and expand it to view all supported AI Service providers.
Choose your Service provider and click on the "plus" button.
Enter a unique service name in the dialog box that appears and click OK.
Your new service name will be displayed in the Service Navigation bar on the left, with its properties on the right.
Select the AI Model you wish to use from the dropdown list.
Click the "Portal Settings" lock button to enter required fields for your Service Provider, which may vary by provider.
Click the "Service Settings" button and set the appropriate model token constraints.
Enter (copy and paste) your provider settings, you need to verify your settings first and then you can press on the "OK" button. to save them.
You also need to assign your new AI Service to an Agent (either a new Agent or by modifying an existing one). Go to the Agents section directly above Services to do that.

AI Service Portal Settings and Model Token Constraints

OpenAI

Azure

Google
- Models
- API Version

Anthropic
- Models
- API Version

VocaGPT includes standard models for each of the AI Services at the time of App release but if the App doesn't know about a new model, you can enter the model identifier as a "Custom" model type. You need to specify your model identifier (see the "Model" links above) in the custom model setting, then the portal settings and token constraints.

See artificialanalysis.ai for a general LLM (model) comparison.

App Operation

Speech recognition is provided by the Apple SFSpeechRecognizer service.
Speech synthesis and associated voices in various languages and genders are all provided by the Apple AVSpeechSynthesizer service.
Additional premium and enhanced voices can be downloaded onto your device by going to: System Settings ⇒ Accessibility ⇒ Spoken Content ⇒ Voices ⇒ your language.

The App Refund Process

To request a refund to Apple, go to App Settings => Info section and press the "Request Refund" button. This will send a refund request to Apple who will adjudicate the request and if granted, the App will receive and process the refund transaction.

App Release History

Version 1.4, August 2024.

Adds new provider: MistralAI with the models: Nemo, Large, Open-8x22B, Open-8x7B and Open-7B Adds new models: Google Flash 1.5

Version 1.3, July 2024.

Adds new models: OpenAI ChatGPT-4o-mini, Anthropic Claude 3.5 Sonnet.

Version 1.2, June 2024.

Adds new models: OpenAI ChatGPT-4o, ChatGPT-4-turbo and Gemini 1.5-pro.

Adds new capability: photo capture and AI request photo attachments to models that support image input (ChatGPT-4o, ChatGPT-4-turbo, Gemini 1.5-pro, Anthropic Claude 3.0 Opus, Sonnet and Hakiu)

Version 1.1, March 2024.

Initial release.