With the recent Beta launch of the voice assistant, HiJiffy continues to revolutionise the way hotels communicate with guests. This latest development within Guest Communications Hub utilises OpenAI’s state-of-the-art technology, including Whisper API and advanced language processing capabilities. The virtual assistant can detect and understand the language of voice clips, providing instant and accurate responses to guests’ inquiries.
This voice assistance solution (or interactive voice response system) is already available on WhatsApp and Facebook Messenger and will soon expand to other channels, such as the hotel website chat widget and other key social media and messaging apps. Guests will be able to easily access the virtual assistant from any device and receive quick and efficient assistance with their needs.
The multi-language capability of the voice assistant
One of the advantages of using Whisper API developed by OpenAI, creators of ChatGPT, enhanced with advanced language processing capabilities, is the ability to understand multiple languages. As a result, HiJiffy’s virtual assistant can communicate with guests in their preferred language, providing a more personalised and human-like experience.
Additionally, the Whisper API’s advanced language detection capabilities will help the virtual assistant understand and respond to complex voice communications, making it more effective in handling guest requests and inquiries.
How Whisper API’s architecture powers HiJiffy’s voice assistant
At a high level, two components within the Whisper API architecture support the voice assistance capacity of HiJiffy’s Guest Communications Hub.
- Speech-to-Text: converting the audio input into text. Using a combination of signal processing and machine learning techniques enables the Whisper API to transcribe spoken words.
- Language Detection: identifying the audio input’s language, enabling the virtual assistant to understand it and respond in the same or the user’s preferred language.
HiJiffy’s proprietary Aplysia OS powers the voice assistant with Natural Language Understanding (NLU) – understanding the meaning and intent behind the words. Deep learning models, such as recurrent neural networks (RNNs) and transformer networks, analyse the text and extract relevant information.
According to a paper titled Robust Speech Recognition via Large-Scale Weak Supervision, written by some of the creators of Whisper AI, it provides transcriptions with significantly fewer errors than previous models. The authors even say that compared to a model developed by Facebook AI in 2020, Whisper AI, on average, makes 55.2% fewer errors than Facebook’s model when tested on 14 different speech datasets. The authors also compared Whisper AI against professional transcriptionists and stated, “Whisper’s English ASR performance is not perfect but very close to human-level accuracy.”
Overall, integrating OpenAI’s Whisper API and advanced language processing capabilities into HiJiffy’s virtual assistant is a significant step forward in developing voice-based communication technologies in the hospitality industry.
Hotels can be empowered to improve guest satisfaction further and streamline operations. This is likely to result in a more efficient and positive experience for guests but also hotel staff, thanks to reduced repetitive tasks.
OpenAI – the team behind ChatGPT
OpenAI is an AI research laboratory that aims to ensure that artificial general intelligence (AGI) benefits all of humanity. It consists of the for-profit OpenAI LP and its parent company, the non-profit OpenAI Inc. Founded in December 2015 by Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, Wojciech Zaremba, and several others.
The company also develops and releases AI models and tools that developers and researchers can use to advance the field. Additionally, OpenAI is also working on developing a range of AI-based products and services that can be used by businesses and individuals, such as the GPT-3 language model and the Whisper API for voice-based communication.
