Can Voice Technology Make You a Better Cook?
I tried three voice-controlled cooking apps, and here’s what I learned
If you’ve read some of our previous articles (VUX buzzwords, Psychology of VUX), you’d know at 55 Minutes, we have a deep curiosity in Voice UX! To take our understanding of this technology to the next level, I decided to do my own research and test out voice-controlled cooking apps.
For some background, I live with my husband and younger sister, and we are all working from home. My husband and I usually take turns to cook quick meals daily and are both “mildly seasoned” cooks (excuse the pun), while my sister, Alanis, well… let’s just say cooking is not her cup of tea.
Following the extremes and mainstreams design theory, I thought “If these voice cooking apps can help improve her cooking experience and skills, they can probably help the rest of us!” So I devised an experiment that tasked her to use pre-assigned cooking apps to prepare lunch for us, for several days; yay, killing two birds with one stone!
3 ways voice can improve your cooking experience
Before sending my sister, Alanis, into the experiment, I briefed her on some of the general things that voice can do to help during cooking. From personal experience, the most helpful thing is that it makes certain tasks hands-free. For example, with Google Assistant on a smart speaker or display, you can:
- Set (and snooze) multiple timers ⏱️⌛
“Hey Google, set a timer for (20 minutes, and an hour)”
“Hey Google, snooze for (two minutes)!” - Find out substitutes for ingredients or tools 🍋🔨
“Hey Google, what can I use instead of (lemon juice)?”
“Hey Google, what can I use instead of (a meat pounder)?” - Convert between different measures* 🥄☕
“Hey Google, how many (tablespoons of sugar is in a cup)?”
Here’s Google’s official guide for other commands.
However, not everyone has a smart speaker or display, and when I tried using Google Assistant on my smartphone, I found it so much more limiting. So I set out to find mobile apps, or Google Actions that I can access from my phone, that had more voice functionalities to help with cooking.
Preliminary Research
Desk Research & Quick Testing
After checking Alanis’ school schedule, I worked out that we can test up to three apps, plus a control. But with so many cooking apps out there, I needed to narrow down the list. I conducted a brief desktop research and considered the apps against the following questions:
To further narrow down the list, I downloaded the eight remaining apps and checked whether the voice functions really work.
One surprising finding was that most of the apps did NOT give any prompts or clues to their voice capabilities — except Voicipe (if we can’t already tell from the name). This meant that users might not even know about the voice features without prior research.
I logged the remaining findings and considerations in this Resource List, but long story short:
Voicipe, Whisk, and Tasty were chosen, and we were ready to get cooking!
Usability Test
Task: Prepare lunch for three tasters (my husband, me, and the cook herself), with help only from the following pre-assigned cooking app/‘assistant’ for each of the four testing days:
Though the task may sound simple, there were actually many variables to consider! For example, here are some known limitations:
- The tester’s cooking skills are likely to improve with each session.
- The tester’s familiarity with using voice input & VUX will likely increase with each session.
- The three tasters will rate the food after, but ratings from the cook herself may be biased 😁
Other factors include our constraints, metrics for success, and measures to account for limitations. If you’re interested to find even more details, you can read my Usability Test Plan.
Key Findings 🔍
As shown by the grey bars in the chart above, the clearest pattern found was that the number of ingredients used steadily increased over the 4 sessions. Each day Alanis was provided with the same number of core ingredients (the top four ingredients in each bar), but she gradually became more adventurous in exploring the kitchen: checking out other things she can use in the fridge, freezer, and cupboards.
To gauge helpfulness and ease of use of the pre-assigned ‘assistants’, we used a 6-point scale (1= very unhelpful/very difficult, 6= very helpful/very easy). The blue bars in the chart above shows that the overall highest scores were tied at 9 points for Day 1’s control and Day 4’s Tasty.
The most helpful assistant for the task was the control (human voice assistant - yay me!), and the easiest to use was Tasty.
The total time was steady at around 50–60 minutes, but ‘learning time’ (time taken to study the app and use it to find a recipe) was by far the longest for Whisk (Day 3) at 10 minutes. The actual learning time might’ve actually been even longer since the user gave on the app!
For the final results of the meals cooked, there were no obvious patterns in the ratings (‘how healthy, hearty, and delicious’), but the cook did consistently give a higher score for ‘deliciousness’ on all four days — fair enough!
Pain points 😩
“I feel like I shouldn’t touch the phone, with my salmonella riddled hands”
- Not 100% hands free — all the apps still required the user to click on the screen whilst cooking. Voicipe was particularly counterintuitive: Alanis had to click on the microphone button every time she wanted to speak, so she still needed to wash her hands to read off the screen.
- Having to repeatedly ask and rephrase questions, as the apps were not able to understand commands that are slightly different from the specific trigger phrases.
- Even after learning the specific commands, Alanis got different results sometimes, e.g. ‘Next step’ occasionally repeated the same step.
- After each ‘Hmm I don’t understand’s, the user could not go back to where she was in the recipe. E.g. Tasty kept bringing her back to the recipes’ home page.
- The apps were not able to answer general / technical questions (also no access Google Assistant’s usual functions e.g. timer, substitution, conversion)
- Some voice functions actually slowed down the process e.g. Whisk read out only one ingredient at a time, waiting for the user to say ‘Next’.
“Reading it is just much faster, no?” — Alanis
Learnings & Opportunities ⭐
“I feel like the app is a bust, so I shall use…my brain”
Alanis’ quote may sound a little harsh, but it perfectly captured the frustrations that she experienced trying to figure out how to use the apps. Through analyzing my observations notes and her comments captured during and after the test, I thought of some ways in which the voice experience can be improved.
Firstly, most recipes are currently written for visual reading. But perhaps a voice-friendly recipe requires a different layout/design! For example, we could provide two formats of ingredients list:
- Concise list— to quickly hear the ingredients when choosing the recipe
- Detailed list — to prep the ingredients whilst cooking: Ingredients and steps grouped according to how they are used.
Secondly, a repeated mistake was calling out “Hey Tasty” or “Hey Whisk” when interacting with the app on Google Assistant. This shows that it makes sense to be able to call on the respective app’s names, so it would be more intuitive to include the additional wake words to the voice interactions. It would also help for the app to provide instructions on how to use the voice functions, such as the different commands to use.
Thirdly, apps that were integrated with a voice assistant (VA) had more functionalities, but were not as intuitive to use. This improvement is likely more to do with the technology, but the VA should still be able to execute its usual functions (like setting a timer) and answer general questions, when accessing the integrated app.
Finally, parts of the cooking experience could be enhanced with visual output in addition to voice. For example, seamless switching between voice and visual interactions such that users can say “Show me how to do the next step” to see videos/photos if the read aloud steps are not clear enough. There should also be an option for the phone screen to stay on whilst in use.
Final thoughts
Whilst writing this reflection, I chanced upon Tiny Chef, previously known as KloveChef (originally eliminated from our list because it wasn’t available on Google PlayStore). Although Tiny Chef is a Google Action and not a mobile app, it’s accessible through Google Assistant on the phone, so I wanted to give it a try too (p.s. it’s also an Alexa Skill).
The verdict: I was pleasantly surprised at how much easier it was to use compared to the other apps! Sure there were still small hiccups (like mishearing ‘radiance’ for ‘ingredients’, and leaving in the middle of our conversation), but the interaction was still particularly memorable because of several delightful surprises!
My favorite was that Tiny Chef played some pleasant music after a step has been read. It also gives the option to go through pantry items and kitchen tools, as well as setting off an automatic timer when the step indicates a time (although it needs you to check back in rather than notifying you when the timer is done).
So from helping you find recipes, choose one, prep ingredients, and actually cook the meal, voice apps and Actions have great potential to create a more seamless kitchen experience using just your smartphone.
Voice-controlled cooking apps alone might not make you a better cook, but can help make you a more confident and adventurous one!
55 Minutes is currently on the lookout for Voice UX projects and collaborations! If you have a project that you’d like to explore using voice UX, or if you’d like to collaborate on a voice project, please drop us a line at hello@55mins.com.
Aurelia is a UX Researcher at 55 Minutes. She studied Medicine and Psychology in London, but her creative side led her to search for a different path! Now, she is in Singapore in a role that utilizes both her critical and lateral thinking skills. Outside of work, you can find her baking, cooking, or hanging out at a cafe, dreaming of future travels.
