Skip to content

Unleash the power of Chatty: the intersection of ChatGPT’s intelligence, DALL·E's creativity, and Whisper's precise audio transcription for your Apple devices with support of 30 languages.

License

Notifications You must be signed in to change notification settings

nicolodiamante/chatty

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

27 Commits
 
 
 
 
 
 

Repository files navigation

chatty

The advanced technologies propelling the latest generation of chatbots have been gaining momentum over the years, revolutionising our world. With the launch of ChatGPT, a competitive scramble amongst the brightest minds has begun in Silicon Valley, each looking to break new ground in what chatbots can do. This shortcut aims to integrate ChatGPT, allowing users to interact with their devices in unparalleled ways. This integration offers users access to three powerful features: ChatGPT's impressive intelligence, DALL·E's innovative creativity, and Whisper's precise transformation of audio into text. This combination of capabilities promises an exceptionally advanced chatbot experience.

Siri + ChatGPT



Integrating ChatGPT is fairly straightforward. However, it requires certain specific prerequisites and links for seamless functioning.

Requirements to get ChatGPT to work with Siri

Get the Shortcut

To initiate, first download the Apple Shortcuts app followed by the Chatty Shortcut. Feel free to rename the shortcut to a title of your preference. However, before you begin using it, ensure that you continue reading for further instructions.



Download the Shortcut


Please make sure to complete the following tasks when transitioning to the new version: Decode the API keys that were previously stored in an encrypted format in the Data Jar. To simplify this process, use the DecodeMe shortcut. Remember to complete this for each key of OpenAI's and ElevenLabs.



Create your OpenAI API Key

To utilise ChatGPT's services with Siri, it is essential to obtain an OpenAI API key. In order to acquire this key, you need to generate a new secret key from your existing OpenAI account. If you don't possess an OpenAI account yet, you can easily create one by following this link. This secret key plays a crucial role in verifying your authentication. Here are the steps on how to do it:

  1. Log in to your OpenAI account.
  2. Search for the ‘Create new secret key’ option and click on it.

Create new secret key

Once you obtain the API Key, incorporating ChatGPT's services with Siri is straightforward. It's important to remember, after making a copy of the key and closing the pop-up, you won't have access to it again. Therefore, store the key securely in a safe location.

OpenAI API Key


How to set up and use ChatGPT with Siri

Setting Up

After downloading the shortcut and copying your API key, follow these steps to put it to use:

  • Begin by opening the Shortcuts app and finding the downloaded Chatty shortcut.

  • Upon activating the ‘Chatty’ shortcut, you'll need to enter your OpenAI API key. You can choose to store this key either directly in the shortcut or in the Data Jar app. However, if you choose to store the key in the shortcut, it's crucial to ensure their privacy. Any inadvertent exposure could compromise your security and potentially result in a breach.

  • Next, continue with the configuration procedure. Follow each essential step attentively to customise the outcome prompt based on your specific preferences for the shortcut. Once finished, you can start using 'Chatty' right away.


Paste your API keys


All set! After completing the integration, you can initiate the shortcut simply by tapping it inside the app or from your Mac's menu bar. If desired, you could also activate the shortcut using Siri. When ready, just say ‘Chatty’ and follow it up with your question or prompt. The shortcut will then generate a response and read it back to you.



chatty launched



Considerations

ChatGPT is one of the foremost chatbots available today, with the capability to generate responses remarkably similar to human interaction. However, like all technology, it does present certain limitations and necessitates particular usage patterns to deliver the expected results. If you've found Chatty less useful, it might be due to the lack of specificity in your input. For Chatty (aka chatGPT) to provide precise and accurate responses, it's critical to clearly define your desired results. General or vague instructions will most likely result in a similarly unfocused answer, while an explicit and well-defined request will generate a more purposeful and useful response.

In summary, to exploit Chatty to its fullest, it is imperative to detail what you want and provide clear directives on what to avoid. Keeping these guidelines in mind, you can harness the full power of this advanced chatbot and simplify your daily tasks.



chatty Output



ChatGPT, DALL·E, and Whisper represent three highly notable breakthroughs in AI technology in recent years. The potential of Chatty lies in combining their capabilities into a single solution, revolutionising how we interact with our devices, and thus significantly simplifying our lives. However, just like with any technological advancement, there are inherent limitations. There may be instances where they fail to produce the anticipated results due to external factors such as server overload or inadequate data. In conclusion, while such technologies undoubtedly simplify our tasks, it is crucial to always consider the input, be mindful of potential limitations, and remember to handle the output with care.



chatty & Whisper Output



What to expect in Chatty

Chatty offers a diverse array of capabilities across its various versions, making it a versatile and valuable AI assistant. With expanded language support, users can seamlessly interact in their native languages, promoting inclusivity and accessibility. Chatty can impersonate up to 16 different roles, including but not limited to roles like Business Strategist, Teacher, Accountant, Career Counselor, Editor, Essay Writer, Interviewer, Mental Health Advisor, Motivational Coach, Personal Trainer, Social Media Manager, and more. Users have the flexibility to customise token usage, enabling precise control over content generation. It provides the convenience of saving and accessing chat history, preserving valuable conversations effortlessly. The integration of Whisper technology enhances Automatic Speech Recognition (ASR), allowing for accurate audio-to-text transcription and voice command-based content creation. Chatty's ability to summarise calendar events, articles, and generate images. Whether engaging in continuous conversations, adapting to multiple languages, or customising system prompts, Chatty is an adaptable AI assistant that continues to evolve to cater to a wide range of user needs, across various domains and applications.


Chatty on Mac



What's new in Chatty

v1.5.3

Release Highlights:

  • New Model Introduction: GPT-4o mini (‘o’ for ‘omni’): The most advanced and affordable model in the small models category. It boasts higher intelligence than gpt-3.5-turbo while maintaining the same speed. GPT-4o mini is designed for smaller tasks. It is recommended to choose GPT-4o mini where gpt-3.5-turbo would have been used previously, as this model is more capable and economical.

v1.5.2

Release Highlights:

  • API Key Storage Enhancement: To resolve issues with script readability, API keys will no longer be encrypted in Data Jar. Previously, encryption caused errors that prevented the Shortcut from reading the API key. Storing keys unencrypted will ensure seamless script functionality.

v1.5.1

Release Highlights:

  • New Model Integration: This update simplifies the model options from 4 to 3. Users now have three distinct models to choose from: ChatGPT-3.5-Turbo, which now points to ChatGPT-3.5-Turbo-0125; ChatGPT-4 has been upgraded to ChatGPT-4-Turbo; and the newly introduced ChatGPT-4o is now available. This adjustment aims to simplify model selection while maintaining advanced capabilities.

  • Enhanced URL Summary Feature: This update refines the function that analyses and condenses web articles into concise summaries when a URL is pasted. The summaries provide a clear overview, capturing about 20-35% of the original text, organised with key bullet points and a final summarising sentence.


v1.5

Release Highlights:

  • Revised Shortcut Structure: The structure of shortcuts within Chatty has been revisited and refined for improved usability and efficiency.

  • New Model Integration: This update introduces two new models for selection. The first, the ChatGPT-4-Turbo model, combines GPT-4's advanced reasoning with enhanced processing speed to ensure quicker and more efficient responses. Additionally, the ChatGPT-3.5-Turbo-1106 variant has been added, offering improved performance compared to the 3.5-Turbo model. This variant is particularly suitable for scenarios requiring slight improvements in response quality whilst still maintaining high efficiency and speed.

  • Updated Prompt Output: The standard prompt and the Do Anything Now (DAN) mode in Chatty have been enhanced for improved continuous conversation support. These updates ensure more precise and user-specific responses.

  • Enhanced Role Prompts: Updated the prompt structure for each role, improving clarity and user engagement, ensuring a more intuitive and efficient interaction experience.

  • Updated to DALL·E 3 with Option for DALL·E 2: Upgraded to the more advanced DALL·E 3 for enhanced image generation capabilities, while still providing users the choice to use DALL·E 2, ensuring flexibility and customisation in image generation preferences.

  • Expanded Language Support: Chatty now offers support for several additional languages, including Ελληνικά (Greek), Hrvatski (Croatian), Norsk Bokmål (Norwegian Bokmål). This enhancement represents a significant step towards inclusivity, enabling a broader range of 30 languages supported by Chatty in their native languages.


v1.4.4

Release Highlights:

  • Prompt Enhancements: Minor improvements to both the standard prompt and the Do Anything Now (DAN) mode to enhance the effectiveness and accuracy of Chatty's responses.

  • Expanded Language Support: Chatty now offers support for several additional languages, including Nederlands (Dutch), Українська (Ukrainian), Română (Romanian), Svenska (Swedish), Suomi (Finnish), Dansk (Danish), and 繁體中文 (Traditional Chinese). This enhancement represents a significant step towards inclusivity, enabling a broader range of users to engage with Chatty in their native languages.

  • Introducing the Business Strategist Role: In this update, the new "Business Strategist" role is introduced within Chatty. This role provides users with a dynamic and insightful approach to business planning and strategy development. Users who choose the Business Strategist role will embark on a comprehensive journey to deeply understand their business objectives and craft effective strategies. This addition empowers users to optimise their business endeavours and make informed decisions, enhancing their overall success and growth.

  • Shortcut for Quick Access: A new shortcut for Mac users has been added: ⇧⌘<. This shortcut simplifies the process of launching Chatty, providing users with a convenient and efficient way to access the services promptly. This improvement aims to enhance user productivity and convenience.


v1.4.3

Release Highlights:

  • Chat History Preservation: A new feature allows users to easily save and access chat history. Whether for reference or convenience, this feature ensures that valuable conversations are securely stored, either in Data Jar or the Shortcut directory. If you opt for Data Jar as your storage option, a dedicated Chatty-Log will be created to house your logs. On the other hand, if Data Jar is not utilised, your chat history will be securely saved in the Shortcut directory, identified with a JSON extension. Rest assured, with this enhancement, you can preserve valuable conversations without any concern of losing them.

  • Alternative Teacher Role: Chatty introduces an ‘Alternative Teacher’ role, offering a unique approach to learning. Users can delve into various subjects, receiving topic-tailored explanations in each session. The option to progress to the next lesson or engage in quizzes fosters an interactive learning experience.


v1.4.2

Release Highlights:

  • Token Customisation: Users gain the ability to customise the number of tokens used. It's important to note that the maximum token limit for content generation is shared between the prompt and completion sections, with the actual limit depending on the model used. On average, each token represents approximately 4 characters in standard English.

v1.4.1

Release Highlights:

  • Language Support Expansion: Chatty adds support for Polski (Polish) and Türkçe (Turkish), enhancing the inclusivity of the user experience.

  • Enhanced Speech Recognition (ASR): With the introduction of the Whisper translation feature, Chatty's ASR functionality is significantly improved. It now enables transcriptions of audio files from supported languages into English text, facilitating convenient voice translation by simply typing or saying ‘Translate’ in supported languages. Chatty efficiently converts audio messages into text, making multilingual communication hassle-free.


v1.4

Release Highlights:

  • URL Summary Feature: To access this feature, copy the URL of the article you are interested in and then type or said the command ‘Summarise’ in any of the supported languages.

    Upon utilising this command, you will receive a concise paragraph consisting of approximately 100 words. Following that, a bullet-point list will highlight the most crucial information, concluding with a one-sentence summary. With this significant improvement, the summarisation feature allows for the quick and efficient extraction of essential content from any article.

  • New Language Support: Chatty now supports English (India) and हिन्दी (Hindi), enhancing its universality. Many thanks for his help to Karthik Jeyakumar in implementing these languages.

  • Automatic Speech Recognition (ASR) with Whisper: Chatty integrates Whisper technology for ASR, rapidly converting audio into text with high precision. This feature allows voice command-based content creation and easy transcription of notes. Whisper allows users to compose content via voice commands, and also provides quick transcriptions of notes and thoughts within Chatty. Users can say ‘Write Down’ in the supported languages for simple transcriptions.

    To activate voice transcription, say the command and Chatty will start recording. Once completed, tap the screen to stop recording and Chatty will transcribe and copy the contents to the clipboard. This feature also enables users to conveniently convert extensive audio messages into text, useful in applications like WhatsApp.

    For example, in Whatsapp:

    • Hover over the message you wish to transcribe.
    • Click ‘Forward’ in the appearing menu.   Forward Icon
    • Choose the individual chat you want to transcript.   Done Icon
    • Click the Share icon (in the right bottom corner).   Share Icon
    • Select Chatty.   Chatty Icon

Chatty Privacy

Users have the right to privacy, which is why Chatty lets you decide which personal data you wish to share during its setup. Chatty is a feature within the Apple ecosystem that offers built-in protection, including situations where third-party service providers work in coordination with Apple. However, Chatty's functionality depends on the OpenAI API. Hence, if you are reluctant about sharing information with a third-party service, you will not be able to utilise Chatty. Should you decide to experiment with Chatty, we strongly encourage you to review the Privacy Policies of OpenAI beforehand.



Shortcut Privacy



Notes

GPT-4

This shortcut is primarily programmed to operate on the GPT-4 model by default. When we compare the operations of GPT-4 model with the GPT-3.5 model, there is little to distinguish between them in terms of performance for relatively straightforward tasks. However, the GPT-4 model outperforms the previous versions, especially when encountering complex reasoning scenarios. If you wish to access the GPT-4-serviced API, it's crucial to note that this privilege is limited to APIs with a history of successful transactions. If an API isn't eligible to run on the GPT-4 model, for instance, if it doesn't meet ChatGPT-4's qualification criteria, it's recommended to default to the GPT-3.5-Turbo model. If you're looking to explore the various subscription packages, you can review the available options here.


DAN Mode

Just as humans have a dark side, so does AI. The Do Anything Now (DAN) Mode empowers Chatty to perform actions beyond its typical AI guidelines, potentially leading to breaches of OpenAI standards, with issues of inaccuracy and offensiveness arising. Consequently, it's advised to use DAN in a thoughtful and responsible manner.


Whisper

The Whisper API has a default file size limit of 25 MB. If your audio file exceeds this limit, you have two options: either break it into chunks of 25 MB or less, or use a compressed audio format. To ensure optimal performance, it is advisable to avoid dividing the audio in the middle of a sentence, as it may lead to the loss of contextual information.

Here are the languages currently supported through both the transcription and translation endpoints: Afrikaans, Arabic, Armenian, Azerbaijani, Belarusian, Bosnian, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Marathi, Maori, Nepali, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tagalog, Tamil, Thai, Turkish, Ukrainian, Urdu, Vietnamese, and Welsh.


Siri

If you come across Siri stating ‘having trouble connecting to the network’ or asking to ‘try again in a little while’, the problem might be due to an unstable Wi-Fi or cellular signal. As Siri does not have an offline mode and operates by sending data to Apple servers, any issue between your device and the server may disrupt Siri's functionality. Also, if the OpenAI servers, which support Chatty (also known as Siri), are overwhelmed, it could potentially compromise Siri's ability to provide lengthy responses.

Try these steps to troubleshoot the issue:

  • Restart your iPhone
  • Check that ‘Listen for Hey Siri’ is enabled
  • Reset Siri to recognise your voice
  • Check the microphone functions correctly

If after these steps the problem persists, another solution is deleting Siri & Dictionary History from the Settings menu. Go to Settings Arrow Siri & Search Arrow Siri & Dictionary History Arrow Delete Siri & Dictionary History.


Language & Region

The Chatty prompt now offers support for 30 languages, including العربية (Arabic), Català (Catalan), 中文 (Chinese), Dansk (Danish), Deutsch (German), Ελληνικά (Greek), English (India, UK, US), Español (Spanish), Français (French), Français Canadien (Canadian French), हिन्दी (Hindi), Hrvatski (Croatian), Italiano (Italian), 日本語 (Japanese), 한국어 (Korean), Nederlands (Dutch), Norsk Bokmål (Norwegian Bokmål), Polski (Polish), Português (Brazil, Portugal), Русский (Russian), Română (Romanian), Suomi (Finnish), Svenska (Swedish), Türkçe (Turkish), Українська (Ukrainian), 繁體中文 (Traditional Chinese), 粤语 (Cantonese).

Settings

The method of configuring the decimal number settings is determined by the region settings of your device. For instance, a decimal number might be set as 0.1 or 0,1.


Resources

Apple

OpenAI

Data Jar

Contribution

Thank you for considering this shortcut. Any contribution to improve chatty's integration with the Apple ecosystem, through feedback, is highly appreciated. Encounter any issues or bugs? Please report them on the issues page.

Nicolò Diamante Portfolio

MIT License