Voice Is Poised to Take a Quantum Leap
Artificial intelligence (AI) and its capabilities have significantly transformed speech technology in a short amount of time. But an equally exciting innovation is anticipated to be an even bigger game changer in the industry: quantum computing. This advancing field in computer science uses the unique principles of quantum mechanics to address problems that lie beyond the capabilities of even the most sophisticated classical computers.
Although still under development, quantum technology is set to overcome complex challenges that supercomputers struggle to solve or cannot solve quickly enough. Quantum computing represents a groundbreaking approach to computation, using quantum mechanics to perform certain tasks much faster than traditional computers. It relies on quantum particles like atoms, ions, or photons to harness the principles of superposition, entanglement, and interference in quantum bits, or qubits. Unlike classical bits used by traditional computers, which are limited to a state of 0 or 1, qubits can exist in multiple states at once due to superposition.
What’s more, entanglement creates a link between qubits, making the state of one qubit dependent on another, enabling parallel processing. When waves interact, they can either combine constructively to form a larger wave or cancel each other out through interference. Quantum computing leverages this interference to amplify correct solutions while suppressing incorrect ones in computational processes.
New quantum chips and systems can operate at a level thousands of times faster than traditional reduced instruction set computers or semiconductor-based CPUs, processing information in a fraction of the time, according to Robert Wakefield-Carl, senior director of innovation architects at TTEC Digital.
“Quantum computing will allow us to solve problems that are currently intractable with classical devices across chemistry, machine learning, finance, and so many other problems we’ve yet to even conceive,” says Bob Coecke, chief scientist and head of quantum-compositional intelligence at Quantinuum.
Dhaval Gajjar, chief technology officer at Textdrip, echoes those thoughts.
“Quantum computing is one of the most exciting frontiers that may change information processing processes for good,” he says. “If you ever wanted to get lost in a maze, a classical computer would check one path at a time. A quantum computer could check all paths simultaneously.”
Indeed, quantum computing is the next major step in the evolution of computing—and it’s bound to have a colossal effect on voice technology.
How Quantum Can Reshape the Industry
Traditional speech recognition uses algorithms to convert audio signals into text, making the process computationally demanding, particularly because of the large amounts of data generated by spoken input. Quantum computers, on the other hand, could significantly accelerate the analysis of these audio streams by simultaneously processing a vast array of possibilities.
“Integration of quantum computing into speech applications offers numerous benefits, which are going to change the face of speech technology,” Gajjar notes. “The most important is the unprecedented processing speed of massive volumes of data. Traditional computing cannot cope with the complexity of human language, including dialects, accents, and contextual meanings. But quantum computers can handle enormous amounts of data and compute several scenarios simultaneously, making speech recognition systems faster and more accurate.”
Consider real-time translation: Current systems often face challenges with speed and accuracy, especially when dealing with complex languages or idiomatic phrases. Quantum-enhanced speech models could simultaneously assess multiple translation options, providing more precise and culturally nuanced results.
“Quantum computing could set a new standard for real-time processing, expanding speech applications across industries such as customer service, healthcare, and entertainment. The technology could make speech interactions more seamless and natural while reducing costs for companies that rely on high-volume voice processing, like call centers or global virtual assistants,” explains Mitchell Cookson, an AI analyst and cofounder of AI Tools.
Quantum computing can also handle multiple states simultaneously and optimize the algorithms behind natural language processing (NLP).
“Speech recognition systems, for instance, rely on recognizing patterns in sound, context, and language structures, which require immense computational power. Current classical methods involve comparing a voice input against massive phonetic and language pattern databases to determine the most accurate match,” Cookson continues. “But quantum algorithms, using qubits in superposition, could compare and match these patterns more efficiently. This would significantly improve the speed and accuracy of transcription, making systems like virtual assistants or automated customer service more responsive and accurate. Quantum algorithms can also improve noise reduction and data compression techniques, making speech recognition more reliable in noisy environments.”
Furthermore, quantum computing could revolutionize machine learning models used for speech synthesis (converting text into realistic speech) and emotion detection (analyzing sentiment and emotional tone in speech). Training deep learning models for these purposes requires substantial computational power because of the large datasets involved. Quantum computers can handle bigger datasets with numerous variables far more quickly than classical systems. This could result in more natural-sounding voice synthesis and more precise emotion detection, enhancing the realism and intuitiveness of human-machine interactions in a variety of industries. And this progress could transform speech technology from basic voice recognition systems into sophisticated conversational agents that truly understand human communication, leading to smoother and more natural interactions.
Current speech technology can identify who is speaking, but quantum computing could significantly improve the precision of recognizing and analyzing individual speech patterns.
“One of the major challenges in improving speech technology today is the recognition of language when a user’s input contains one of potentially hundreds of accents or dialects,” Sam Danby, head of voice at Boost.ai, says. “The integration of quantum computing directly addresses this challenge by improving the speed and accuracy of existing NLU models, which is essential to maintaining user satisfaction.”
Envision a personal assistant that can understand and follow commands and detect the user’s mood by the tone. Using quantum computing, this assistant could quickly realize the subliminal messages of the user’s speech and provide well-tailored responses.
“Think of how you would inquire about your daily schedule from a virtual assistant with a worried tone. A quantum-enhanced system would recognize the stress involved and might respond with more empathy or even suggest taking a break, turning a routine interaction into a nicer experience,” Gajjar says.
Quantum computing can also dramatically reduce the training time for speech and language models.
“The training time for large language model (LLM) GPT-4 was around five to six months. It’s likely that quantum computing would at least halve the training time required for speech recognition and text-to-speech (TTS) training models,” says Deborah Dahl, principal of Conversational Technologies. “This training would happen behind the scenes, while the speech systems are being developed, and wouldn’t be apparent to users except in the form of more accurate speech recognition and better TTS. Faster training times would greatly reduce costs, reduce energy footprints, make it possible to use larger datasets, make some applications practical that are currently not possible, and make it more realistic for smaller companies to train their own models.”
Coecke also points to amazing progress lately in a new field called quantum natural language processing (QNLP)—which exploits the common structure between quantum theory and language.
“Recent work has shown advantages relative to modern LLMs driven by AI, such as interpretability and compositional generalization,” Coecke says. “More specifically, language can naturally be treated as a quantum process and, as a result, is subject to quantum computational benefits. Particular features of current QNLP include language-neutral but still fully interpretable representations—meaning-awareness well beyond current LLMs, which can substantially help with speech recognition—and structures that are intrinsically multimodal and can strongly support multimodal forms of speech technology.”
Enticing Hypotheticals
The prospects and possibilities for speech technology benefiting from quantum computing in the real world are tantalizing. Imagine, for instance, a real-time translation app on your smartphone that goes beyond merely transcribing spoken language; it comprehends context, tone, intent, and even emotion.
“We could have virtual assistants that will sound so humanlike, you will not be able to distinguish them from a human and can use natural language understanding based on LLMs that will be able to understand anything said by the caller in practically any language. Real-time voice translation will allow agents to service customers in any language, no matter what they or the customer speaks, meaning you can staff contact centers with fewer people and fewer requirements on agent language skills,” Wakefield-Carl says.
In international business meetings, quantum-enhanced speech technology could effortlessly eliminate language barriers, enabling smooth conversations among participants from various countries while preserving meaning and nuance. This ability is particularly vital in high-stakes scenarios, such as legal discussions, where accuracy is essential.
“Or consider voice diagnostics in the medical field,” explains Waseem Mirza, presenter of #TheFutureTECHShow and an independent tech consultant and analyst. “Case in point: Subtle changes in speech patterns could indicate early neurological conditions. Quantum-enhanced algorithms may detect these with heightened sensitivity, leading to earlier intervention and more personalized healthcare solutions from neurologists.”
Conventional speech recognition systems often find it challenging to accurately interpret language in noisy or chaotic settings. However, quantum computing could significantly improve noise-filtering algorithms by simultaneously processing a large number of potential signal variations.
“This would allow for cleaner and more accurate real-time speech recognition, even in environments like busy airports or concerts,” Cookson states. “For example, customer service agents equipped with quantum-enhanced speech systems could handle calls from noisy backgrounds without missing critical information.”
Imagine a quantum-powered documentation app for doctors that could easily analyze multiple layers of audio inputs in real time and better separate speech from noise. Or visualize a virtual language tutor run by quantum technology.
“This app, teaching a second language to users, could immediately change its teaching methods based on the speech pattern of the student, how quickly one learns, and how emotionally sensitive that person is,” Gajjar says. “For example, if a student pronounces something badly, the system can have a real-time analysis of their speech against millions of native speakers’ pronunciations and give immediate feedback and personalized exercises to improve on that specific condition of their speech.”
Key Players
Many of the major names involved in quantum research and development are obvious. But lesser-known organizations are also playing a part. Here are some of the important movers and shakers in this space, according to the pros:
IBM is the most significant name and a true pioneer in quantum computing over the years. “IBM’s Quantum Experience platform not only allows researchers access to crucial processors but also supports several projects aimed at determining how quantum algorithms can be put into use to enhance machine learning and natural language processing,” Gajjar says. “These researchers are actively investigating how applying quantum computing can optimize neural networks and enhance the accuracy of speech recognition systems.”
Google is another giant stirring the pot. Its Quantum AI division is developing new methods to enhance machines’ abilities in speech and language; with advancements in quantum algorithms, it aims to create real-time translation services that would allow people to communicate across languages seamlessly.
Quantinuum focuses on QNLP and could play a pivotal role in future speech technologies. The business has been a pioneer in quantum AI models that are interpretable by design, bridging quantum mechanics and linguistic structures and aiming to revolutionize how machines understand and process human language. Such models could play a central role in future speech technologies.
Rigetti Computing specializes in quantum hardware and software. “Their work on hybrid quantum-classical algorithms can significantly enhance the performance of speech technologies,” Gajjar says. “They’re developing tools that could make speech recognition systems less susceptible to variations of accent and dialect.”
Academic institutions, especially MIT and Stanford University, continue to conduct groundbreaking research at the intersection of quantum computing and linguistics. Researchers there are investigating quantum machine learning techniques to enhance speech synthesis systems.
Government agencies/organizations, such as the U.S. Defense Advanced Research Projects Agency (DARPA), the National Institute of Standards and Technology (NIST), the European Union, NASA, and the University Sciences Research Association (USRA).
Other industry disruptors mentioned by the experts include Genesys, NICE, Microsoft, Amazon Web Service (AWS), Meta, and D-Wave Systems.
What Needs to Happen Next
While the future looks bright, several key developments are needed before quantum’s true potential for advancing the speech tech market is realized.
“For example, better algorithms tuned to speech processing applications will be required since most algorithms exist largely on paper today,” notes Gajjar. “Error correction techniques are also important because quantum systems are fragile. And more effective hybrid quantum-classical models are needed to bridge the existing technologies with newer capabilities in quantum to provide a much easier path for integration,” says Gajjar, who believes we are at least five years away from major quantum advancements in speech technology that will be used by businesses and consumers.
More importantly, quantum hardware must become more stable and error-resistant, as current systems suffer from quantum noise that hampers reliability.
“One of the biggest challenges is that the physical qubits are very sensitive to disturbances from factors such as noise and temperature, which could lead to loss of information. Solutions have to be found to increase the robustness of quantum computing,” Dahl cautions. “Another barrier is that quantum computers are very expensive right now. It’s estimated that a single qubit costs $10,000 and a practical quantum computer could cost tens of billions of dollars. Only the largest and most well-funded organizations or consortia can afford those kinds of costs.”
Wakefield-Carl agrees that the expenses of setting up and running a quantum computer will be a detriment to niche players.
“However, I don’t feel that individual computers will be required, as broadband connections and personal device connectivity will allow access to needed resources and services without the need for individual investment. There will most likely not be suppliers of quantum computing services, but applications based on the services run by the original players.”
One approach to lowering costs is to offer quantum computing resources through the cloud, allowing most organizations to avoid the need to host quantum computers on-site.
Another issue that needs attention is the scalability of quantum systems. Current quantum computers lack enough qubits or sufficient coherence time to manage the large datasets needed for speech applications like automatic speech recognition or NLP.
“We also need to develop quantum algorithms specifically designed for speech technology tasks, such as real-time translation and voice recognition,” Cookson says. “In the near term, however, hybrid quantum-classical systems could bridge the gap, allowing quantum computers to handle complex optimization problems while classical systems manage language processing.”
Looking Ahead
Despite these hurdles, industry insiders are optimistic that quantum is on the fast track in the voice field.
“Companies are already conducting experiments on how quantum machine learning algorithms can improve speech recognition accuracy and efficiency,” says Gajjar. “It is really difficult to outline a timeline for when this innovation will be there, but the advancements being demonstrated in quantum computing technology and its application to speech technology are fast-paced.”
Gajjar insists that the promise this convergence holds is gargantuan, and “with the excitement building over new quantum developments, the wait for meaningful integrations may not be so long after all.”
Mirza is also encouraged by the accelerated pace of progress observed so far and quantum’s ability to completely alter the speech tech landscape.
“It has the potential to democratize access to advanced speech technology, making high-caliber capabilities available to a broader range of users and smaller companies,” Mirza adds. “As this technology evolves, it could unlock innovations that transform how we interact with machines, making voice-driven tech more intuitive, adaptive, and capable than ever before.”
While it’s true that advancement is currently slow, don’t forget the lessons learned since the advent of generative AI.
“The innovation curve of a particular technology can change practically overnight,” Danby says.
Erik J. Martin is a Chicago area-based freelance writer and public relations expert whose articles have been featured in AARP The Magazine, Reader’s Digest, The Costco Connection, and other publications. He often writes on topics related to real estate, business, technology, healthcare, insurance, and entertainment. He also publishes several blogs, including martinspiration.com and cineversegroup.com.