The Perfect Retailer Loyalty Program App
10 minutes
Back to Blog
Business
The future of voice commerce: how to use speech recognition and voice search in 2025
Mar 7, 2025
12 minutes
Contents
While building an immersive customer experience today, don't limit your offerings to text-based communication: today's customers want to be able to talk directly to their favorite brands via voice search and speech recognition. According to a PYMNTS Intelligence report, 60% of U.S. consumers regularly use these technologies.
By 2023, only 28% of small and medium-sized businesses will have implemented voice search and speech recognition in their services. Today, 85% of companies expect to see widespread adoption within the next five years, and 66% see voice-enabled experiences as critical to their future business strategies.
Driving this interest is the ability to quickly identify the user and "guess" their needs. A voice assistant can authenticate a customer, instantly access their previous orders, and make a personalized offer. And it can even recognize moods to find the right tone and arguments for communication. At the same time, errors in data interpretation are still holding the technology back. In this article, we look at success stories, integration challenges, and failures.
We recently included voice-activated ordering in a review of restaurant and retail trends for 2025. This inspired us to create a separate guide to help businesses understand the solution and take the first steps toward implementation.
Here you’ll learn:
- What is voice commerce?
- How does it benefit the business?
- How is voice commerce changing modern retail?
- What are some of the successful examples of voice commerce and the failures in the use of the technology?
- How has the technology changed with the development of AI?
What is voice commerce
Voice commerce (or v-commerce) is a shopping approach using customers' voice instead of typing or clicking. It allows users to search for products, place orders, and make purchases simply by speaking to a voice assistant, integrated in the device. To accept a voice request, systems need microphones that can pick up the data. This is why most developers are focusing specifically on software for mobile devices, as well as more than 1 in 4 people regularly use voice search for online shopping on their smartphones.
The technology is represented by two directions:
- Speech recognition – data processing that focuses on recognising spoken words. In simple terms, 'what was said';
- Voice recognition – identifying the personality of the speaker who made the request. In simple terms, 'who said it';
According to Statista, the number of digital voice assistants in use will reach 8.4 billion units by the end of 2025. This trend is closely related to the development of AI, which has changed the way the world understands working with this technology by improving its ability to understand complex queries and respond in a more human-like way.
How do speech recognition work
Today, speech processing includes the following steps:
- Audio input – a microphone captures the audio and then converts the sound vibrations into electrical signals;
- Data pre-processing – the audio signal is cleaned using Mel Frequency Cepstral Coefficients (MFCCs) to remove noise and normalize to efficiently represent the audio signal. This allows the system to focus on relevant patterns and digitize them into a format that can be processed by a computer, which is essential for further analysis;
- Feature extraction – the system analyzes the audio to extract key features such as pitch, tone and frequency;
- Pattern recognition – scanning the speech and matching it to the desired voice fingerprint. AI analyzes the message and the speaker's tone and other distinctive features;
- Speech processing – the recognized patterns are converted to text and natural language processing (NLP) algorithms interpret the meaning.
✍ The result of speech recognition should be a targeted action in response to the user's request. For example, to place an order, to perform an authorization, to provide customer support.
The benefits of voice commerce
According to Data Reportal statistics, approximately 20.5% of people use voice search, and 8.4 billion voice assistants are expected to be in use worldwide by 2025. The reasons why this technology is in demand now are based on the following benefits:
- Speed and accuracy – modern AI-powered speech recognition systems achieve 90-95% accuracy and provide instant responses;
- Accessibility and inclusivity – speech-to-text enables hearing-impaired users and people with disabilities or mobility issues to communicate with businesses more easily;
- Hands-free and multitasking capabilities – the solution is ideal for providing interaction with service while driving, cooking, doing workouts and other situations where typing is inconvenient or impossible;
- Multi-language support – enables businesses to reach global audiences and enter new markets with native offerings and a deep understanding of the needs of the target audience by collecting data via voice patterns during the ordering process.
- Availability – voice AI ordering systems can take phone or app orders even during off-hours without the need for human staff.
All of the requirements described and the solutions offered by the technology are in line with the trend towards offline and online integration. The ability to communicate with the customer at any time increases the conversion rate from initial inquiry to actual order by 2 to 3 times. This is why voice commerce is increasingly being used in foodtech and retail, where the need to buy can be spontaneous and immediate.
How to use voice commerce tools
According to the Business Research Company, the global voice commerce market is expected to reach approximately $151.39 billion in 2025, and over 60% of smartphone users use voice search when shopping. Speech recognition technology has become an integral part of both restaurant apps and retail websites, providing innovative solutions and improving the customer experience. Look at the impact of the solution on the business.
Automated drive-thru and voice ordering
AI-powered voice bots allow to automate voice-based order taking and customer inquiries in restaurant apps. These systems use natural language processing (NLP) and AI-powered speech recognition to take orders, answer common questions, and route calls without human intervention. This helps businesses manage high call volumes while improving efficiency and customer satisfaction.
E-commerce marketing trends show that AI-based assistants and voice search enable personalized experiences by using speech recognition to analyze past orders, preferences, and real-time suggestions. These systems integrate with self-service kiosks and other device systems, allowing customers to place orders effortlessly.
One of the most famous and viral examples was a collaboration between Domino's and Amazon Echo. Customers can change the order of their favorite dishes by simply speaking to the Amazon Echo, and the system immediately sends the updated order to the nearest Domino's. And the speaker will notify them of the change in delivery status. As the result, Domino's has seen over 60% of orders being placed digitally, streamlining the customer experience.
And here's our case study on how we implemented real-time streaming order status tracking for a fattening food delivery company.
Yapoki: mobile delivery app for the future of the Enterprise
Here we’d like to share our experience in developing a delivery application with big ambitions and complex architecture. It all started with the customer's dream - make it like Enterprise (the direct reference was Dodo Pizza).
Another enthusiastic company actively using speech recognition is Wendy's. As of the beginning of 2025, the company has already equipped about 100 drive-thrus with voice-enabled AI and spent more than $53 million on digital services. However, not all customers are happy with these changes. Some find that the voice assistant often interrupts them while they are placing an order, and can also switch off when you pause.
For the same reason, McDonald’s is reportedly planning to end its AI-powered automated drive-thru ordering experiment at more than 100 locations after angry customers reported receiving items they didn't order. The technology has had numerous glitches, most notably misinterpreting customer orders to sometimes hilarious effect.
Customer support
Marks & Spencer recognized the need to modernize their customer service by implementing a centralized, cloud-based contact center. To improve the customer experience, they wanted to replace their legacy system with AI-driven automation for more accurate, scalable, and efficient call routing. These improvements led to the following results:
- 90% call routing accuracy;
- 98% caller retention and response rate;
- 10 seconds saved per contact center call.
The market for AI restaurant voice assistants to replace traditional restaurant hosts is also growing. One such example is AI-host Jasmine, who "works" for Bodega, a restaurant in San Francisco. She handles customer calls and can answer questions such as table availability and making pre-ordering. And her tone and emotion can be customized, depending on the nature of the conversation.
Inventory management
Restaurants and retailers can now track and manage inventory using voice commands through AI-powered systems. Employees can simply speak into a voice-enabled device to check stock levels, update inventory records and generate reports, reducing manual effort and minimizing errors. This streamlines the inventory process, making it faster and more efficient while preventing out-of-stocks or over-orders.
Voice shopping with smart assistants
Customers can add items to their shopping carts, track their order status and complete purchases using simple voice commands. These systems work by combining a user's shopping history and preferences with speech recognition technology to provide a seamless, hands-free shopping experience.
Walmart has developed an end-to-end machine learning system to personalize conversational voice shopping for its customers. This system integrates with platforms such as Google Assistant and Siri, allowing users to engage in voice-based shopping experiences. By analyzing customer preferences and behaviors, Walmart's solution provides tailored product recommendations, streamlines the shopping process, and improves overall customer satisfaction.
Multilingual voice commerce
Multilingual voice assistants enable businesses to expand their customer base and serve different markets. These AI-powered voice systems use speech recognition models trained in different languages to understand and process commands from non-English speakers, making online and in-store shopping more accessible.
Amazon's Alexa offers a multilingual mode, allowing users to seamlessly interact with the assistant in multiple languages. For example, users can communicate with Alexa in English and Spanish interchangeably, allowing businesses to engage with a diverse customer base without language barriers. This feature increases accessibility and inclusivity in voice commerce.
Frauds recognition
In addition to receiving and processing data, speech recognition can also be used for deep analysis to find speech patterns characteristic of criminals. In the financial sector, AI is already being used to sift through huge volumes of communications, reducing the number of false positives compared to traditional 'trigger word' search systems.
For example, speech recognition systems are already being used to monitor the communications of traders on Wall Street and in London to detect financial crime. AI learns from data from regulations, financial reports and regulatory communications and can translate jargon into understandable language. For example, the system can recognize coded messages such as 'Ca11 m3 n0w' or even encryption using emoji.
On the other hand, it's important to remember that many AI models are capable of faking a person's voice, so you may want to think about additional options for completing an order or check-in, such as secret questions and decentralized authenticators.
Challenges and pitfalls
Despite its growing popularity and trends, voice commerce technology still faces challenges such as privacy concerns, speech recognition accuracy and the need for competitive differentiation.
AI "hallucinations"
A case where systems generate fabricated or inaccurate responses. For example, researchers have found that Whisper is prone to hallucinations, in which it fabricates text or entire sentences that are not present in the original audio. These hallucinations have included violent rhetoric and fictitious medical treatments. Ensuring accuracy and reliability remains a critical focus in the development of speech recognition systems.
Multilingual & dialect misinterpretation
Despite its growing popularity and trends, voice commerce technology still faces challenges such as privacy concerns, speech recognition accuracy and the need for competitive differentiation.
WebView limitations
Speech recognition systems often struggle to accurately interpret different accents and dialects. The research has shown that New York City, New Jersey and Long Island accents are among the most difficult for AI voice recognition systems to understand. Users with these accents often experience misinterpretations, leading to frustration and decreased trust in the technology.
Privacy & security concerns
The collection and use of voice data raises significant privacy issues. During a Senate committee hearing, Amazon executives were unable to specify the number of voice recordings collected by Alexa devices, raising concerns about data use and consent. Such ambiguity can undermine user trust in voice-activated technologies.
Tech shortcomings & User frustration
Despite advancements, Amazon's Alexa has not significantly evolved in functionality, which often leads to user dissatisfaction. Many users find that voice assistants can't perform more than basic tasks, and increased functionality often requires specific commands and troubleshooting, making the experience cumbersome.
Voice technology development trends
Let's take a look at the major changes in speech processing and the trends that are shaping its value today.
From Speech-To-Text to Speech-To-Speech
Previously, speech processing was based on speech-to-text conversion, where the data is converted into text or commands that can be executed by the system, thus completing the recognition process. However, this process was quite lengthy, while users expected a faster response from the system.
But after releasing Speech-to-Speech technology it's getting easier to facilitate real-time communication between humans and machines. Speech-to-speech combines speech recognition, natural language processing (NLP) and voice synthesis to understand and respond like a real person. This technology can convert spoken words into different voices, potentially breaking down language barriers.
The main difference with Speech-to-Text is that Speech-to-Speech technology also uses AI and machine learning models to accurately identify and transcribe different accents, dialects and speech patterns. The workflow is as follows:
- Automatic speech recognition (ASR) accurately identifies and transcribes different accents, dialects and speech patterns using AI and machine learning models/.
- The machine translation algorithm processes the text and translates it into the target language.
- Natural language processing (NLP) analyses and interprets the transcribed text for context and meaning.
- Speech synthesis converts the translated text back into spoken words in the target language. Text-to-speech (TTS) technology converts written words into audible speech, using AI and machine learning models to generate natural-sounding speech with variations in pitch, rate, pronunciation and inflection.
With such a solution, you'll be able to select any voice for a response by sending the next prompt. And enhanced MFCC techniques allow the system to pause, allowing users to clarify the nature of their query and have an immersive experience with the voice assistant.
AI-driven voice technology
AI has become a game-changer for improving voice recognition accuracy in food tech, particularly in restaurant ordering systems and drive-through services. Here are the key ways AI is improving voice recognition accuracy:
- Noise reduction – advanced AI techniques help filter out background noise in busy restaurant environments, ensuring clear and accurate order capture;
- Personalization – AI systems can remember customer preferences and previous orders, reducing the likelihood of misunderstandings and improving order accuracy for repeat customers;
- Integration with other systems – AI voice recognition works seamlessly with other restaurant technologies, such as digital menu boards and kitchen display systems, minimizing errors in order transmission and preparation.
By leveraging these AI capabilities, foodtech companies have significantly improved the accuracy of voice recognition systems.
Generative AI and humanized voice assistants
Recent advances in AI-powered speech recognition systems have significantly improved their capabilities, resulting in more natural and intuitive user interactions. Key developments in 2025 relate to the integration of generative AI into voice assistants:
- Amazon's Alexa Overhaul – Amazon is preparing to relaunch its Alexa voice assistant as an AI 'agent' capable of performing practical tasks and acting as a comprehensive concierge service;
- Google's Gemini Live – Google is bringing the latest iteration of its voice assistant, Gemini Live, to iPhones. This voice-based feature allows users to have natural conversations with the chatbot, offering functionalities such as interview practice, travel advice and creative brainstorming;
- OpenAI's Whisper – Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual and multitasked data collected from the web. This extensive training results in improved robustness to accents, background noise and technical language, enabling transcription in multiple languages and translation into English.
Olga V.
Business Manager
Get a free consultation on using AI solutions for your business
Another trend in speech recognition development is related to the demand for multimodal and emotionally aware AI models. A review of 42 studies from the last decade, organized by Hindawi, shows that there is a lot of research on making chatbots more emotional. About 57% of the studies use advanced models to make the chatbot say things that sound like emotions.
And here are the latest solutions aimed at humanizing voice assistants:
- FunAudioLLM – this family of models is designed to enhance natural speech interactions between humans and large language models (LLMs). It includes SenseVoice for multilingual speech recognition, emotion recognition and audio event detection, and CosyVoice for natural speech generation with control over multiple languages, timbre, speaking style and speaker identity;
- Meta's AI chatbot with celebrity voices – Meta launched an AI chatbot with voice interaction capabilities, including celebrity voices such as Judi Dench. This development aims to provide a more engaging and personalized user experience.
To use or not to use
Voice commerce is no longer a futuristic concept – it's an active transformation shaping the retail and hospitality industries. But it's still bringing both benefits and pitfalls for all who will use it:
- Businesses – adopting voice AI is no longer optional, so businesses must optimize for multilingual, personalized and AI-powered voice experiences to remain competitive;
- Consumers – convenience, hands-free interactions and personalization will drive adoption of voice commerce, especially for everyday purchases;
- Developers – AI accuracy, privacy concerns, and hybrid AI-human interactions must be addressed to fully realize the potential of voice commerce.
By 2030, this solution will be an integral part of the digital experience, transforming the way we shop, order food, and interact with brands. Companies that innovate early will reap the rewards of seamless, AI-powered commerce.
FAQ
What is voice-based eCommerce?
Voice-based eCommerce refers to the process of using voice commands through smart assistants such as Amazon Alexa, Google Assistant, or Apple Siri to search for products, compare prices, and complete online purchases. This hands-free shopping experience allows users to interact with online stores using speech recognition technology.
What is an example of a voice search?
An example of a voice search would be saying, “Hey Google, where can I buy running shoes near me?” or “Alexa, order a pack of batteries from Amazon.” These queries use speech recognition to retrieve relevant search results or process an order.
How is voice search revolutionizing the e-commerce shopping experience?
Voice search is transforming eCommerce by making shopping more accessible, convenient, and faster. Users can search for products, check prices, and even complete purchases without typing. Retailers are integrating voice assistants to provide personalized recommendations, streamline checkout processes, and enhance customer engagement. This technology also enables a more natural and intuitive shopping experience.
How do I get my business on voice search?
To optimize your business for voice search:
- Make sure your website is mobile friendly and optimized for fast loading speeds.
- Use conversational and long-tail keywords in your content.
- List your business on platforms such as Google My Business and other relevant directories.
- Implement structured data (schema markup) to help search engines understand your content.
- Optimize your product descriptions and FAQs for voice-friendly queries.
- Consider integrating voice commerce capabilities into your online store.
Is voice search on the rise?
When we talk about e-commerce marketing trends in voice search, it's a tool that more and more consumers are using in their daily lives to find new products and services. If companies ignore voice search, they are ignoring a new revenue stream. That's why voice search will become a new standard for digital marketing.
What is the impact of voice search on digital marketing in 2025?
In 2025, voice search is expected to play a crucial role in digital marketing by influencing SEO strategies, content creation, and ad targeting. Marketers will need to focus on:
- Optimize for conversational and intent-based queries.
- Improve local SEO, as many voice searches are location-based.
- Create voice-friendly content, such as FAQs and short answers.
- Leverage AI-powered personalization to provide relevant product recommendations.
- Explore voice-enabled advertising opportunities to capture voice-first consumers.
With these trends, companies that adapt early to voice search will gain a competitive advantage in reaching their audiences more effectively.
You may also like:
