I've been getting more involved in kitchen activities lately. The COVID-19 pandemic has urged many to get their hands dirty while making banana bread, cinnamon buns, Dalgona coffee... the trends keep changing! Being at home for so long brought trivial details and activities around the house under the spotlight. One of these trivial details was the pressure cooker whistles which can be heard in every Indian neighborhood.

We've used traditional pressure cookers since I was a kid, it's been part of my growing years. For those who don't know what a traditional whistle cooker is, it's a pressure cooker with a whistle which indicates whether peak pressure has reached. Although many shifted to electric cookers, we stuck to it, and like most Indian families we count the number of whistles to know if the food was cooked. But even after all these years listening to cooker whistles at home, it still feels annoying when mom asks me to keep a lookout for the cooker whistles, and it has become more apparent during the lockdown. I constantly found myself running around the house asking people if they heard any whistles, even burning the rice couple of times. But it was different this time. It wasn't a trivial problem anymore, now that my design instincts had kicked in. I knew I can't be the only one facing these issues! So I set out on a journey to solve the infamous whistle conundrum.

🔬 research study

In order to validate my hypothesis, I put out a survey to understand rice cooking methods adopted in Indian kitchens. This was the first part of my mixed-method approach towards understanding my stakeholders. The results were in line with my assumptions, with some interesting findings.

bar plot of rice and pulses cooking methods

Majority of Indian households still use traditional whistle cookers, with a surprisingly decent adoption of electric cookers. But what really shook me was that people still used regular open utensils to cook rice. Below is a Sankey Plot which breaks down people's experiences with their cooking methods.

Sankey plot of cooking practices

We see that 54% of the people who used traditional whistle cookers found it problematic, in terms of overcooking, undercooking or burning the rice because they missed the cooker whistles. Something to note is that the survey was filled by young adults who are 20-27 years old. Most of them were students who are currently studying or recently employed, and don't regularly involve in kitchen activities, which explains why most respondents found it problematic, because one would assume that years of cooking rice in pressure cookers would've tempered the learning for most house cooks.

The survey helped me set up interviews with a few participants who found the entire affair troublesome, to better understand the problems they face. I also talked to household members of 4 of my participants to get their perspectives.

🔭 findings

---

The interviews and the survey data gave me some really interesting insights.

• The duration of the whistle is important. There are times when the whistle comes for a jiffy, but this is something which is not counted towards the tally.
• The responsibility of looking after the pressure cooker is usually given to someone when the cook has some other work. For example, if mom has to get ready for work or a zoom meeting, she will ask her daughter to keep a look out for the cooker whistles, who reluctantly agrees, but forgets.
• In independent houses and villas, with kitchens being on the ground floor and rooms being usually on the next floor, it's almost impossible to hear the whistles. This forces family members to come down and sit somewhere near the kitchen and continue their work if they are asked to switch off the cooker after a few whistles.
• People said they would rather continue working on their assignment submission, continue their yoga session, watching TV or binging Netflix if they weren't asked to count the whistles.
• Indian families do not trust the pasteurization process which commercially produced milk goes through, and boil the milk before using it. When asked if there are other problems they face in the kitchen, many survey respondents mentioned they forget to switch off the gas stove and the milk almost always either overflows or burns. Although quite relevant, it's a problem to solve for another day.

Finally, the running narrative among all the participants was that they don't like standing in the kitchen and waiting for the whistles. I created two personas based on these findings, which helped me design for my users better.

Personas created based on findings

As a final step, I decided to conduct a literature review to delve into the nitty-gritty of pressure cookers, their history in India, and what is considered to be the best way to use them.

📖 literature Review

Pressure cookers' history¹ can be traced back to 1910, with a tiffin carrier called icmic cooker, made with coal and water at the bottom of a tall compartmentalized chamber to act as a slow cooker. Many years later Automa and Prestige took over the traditional cooker market with their pressure cookers which got very popular in Indian households. In his book, "How TTK Prestige Became a Billion-Dollar Company", TT Jagannathan explained how Indian women were happy with the reduced fuel consumption and time spent sweating in the kitchen.

How people use their pressure cookers varies from household to household, something elucidated through the survey. When conducting my user study, I was asked to check out two discussion threads (here and here) on a Facebook group called, "Simple recipes for complicated times". The first thread started with a discussion on how counting whistles to decide whether your food is cooked is wrong because you lose all the pressure built up in the cooker, wasting energy in this process. The correct way of doing it is by keeping the pressure inside, by putting the flame on high, letting the pressure build, then putting it on simmer, and using a timer to time the cooking process. People commented on the post that they never knew this and are grateful to learn the right way. I was puzzled to learn that I've been using pressure cookers the wrong way all this time, and immediately pulled up manuals for popular pressure cookers used in India, which vehemently pointed out how wrong I've been. Three days later a second thread was made which talked about how even though counting the whistles is the wrong way of going about cooking, it's a convenient way for the cooker to tell you that the food is cooked. The post was well-received, with a lot of positive comments from people who felt relieved to know they weren't alone.

In a world that gets busier by the day, our kitchens get messy and complex. Family members have to keep up with their morning schedule, their kid's lunch for school, making sure the boiling milk doesn't overflow, and getting ready for work. In a world like this, the cooker whistle quickly becomes a vital affordance for indicating that the food is cooked. Even if it's not the right way of using pressure cookers, it has worked well for many households. But it is still problematic for some, especially the younger generations, and they are the people I am designing for.

Although I was convinced that counting cooker whistles was an imperfect but convenient way of using pressure cookers I was a little concerned about the convenience-energy tradeoff. So I decided to conduct a little experiment to find out how much energy do we end up wasting by letting the whistles go off.

🧪 experimentation

I did some math and calculate the energy consumptions for the top two cooking methods my participants practiced,

1. The correct way of using pressure cookers, where you build up the pressure by keeping it on high heat and then putting it on simmer to let the steam cook the food.
2. The relatively convenient way of just letting the cooker do its thing and count a set number of whistles based on the type of food.

The idea was to see if there is a significant difference in energy consumptions in both these techniques. I used an induction cooktop to measure the energy used for each method, and repeated the experiment 2-3 times for both methods to get an average estimate. The below infographic goes into the results of my experiments.

Plot of experiment results

The plot shows the experiment conducted for different power levels, and it tells us that the energy consumption is almost the same for both the methods, which means you don't save much energy by doing it the right way. This only tells us that forcing people to change their mental models about using pressure cookers isn't wise, and the correct path to solving this is to design a solution which will help the younger generation Indians count the cooker whistles easily.

⚙️ prototype development

There are two core components to the prototype development

1. Seamless recognition of the cooker whistles
2. Notifying the user when the whistles have been counted

For the first part, the findings from the research study helped me identify various activities which the users engage in and would engage in if they weren't consciously focusing on a task as mundane as looking out for cooker whistles. These activities laid out a framework for the data collection process, required to train a robust recognition model which can accurately detect and count the cooker whistles.

The data collection took around a month, where I used my smartphone's microphone to record cooker whistles, while emulating the settings in which the user will be using the app. The recordings were taken in a variety of surroundings and situations based on my findings.

• Noises from the kitchen utensils, running water, ice dispenser, cooking, environmental noises, television, speakers, conversations and keyboard clicking.
• In surroundings like kitchen, main hall, bedrooms located far away from the kitchen

This was done while considering various distances between the microphone and the pressure cooker while recording. The audio files were then manually parsed to extract audio snippets, labelled as "Whistle" and "Noise", and standardised to 16kHz sampling rate. For model building, I used the createML framework's sound classification template to create a robust yet small-sized Neural Network (at just 4.6MB!) which was able to classify cooker whistles with 99% accuracy! The model further checks if the whistle sound is detected for at least 2 seconds to count it as a whistle.

For the second part, I brainstormed all possible scenarios where the app will be used based on the personas developed and how the app can help.

1. The user is busy with work on their computer - People working on their computers are relatively more attentive to mail notifications than their smartphone notifications, and in such cases, they can opt for a email notification and get back to their work.
2. The user is in a room far away from the kitchen - People can opt for an alarm notification and leave their smartphone somewhere close enough to the kitchen but also close enough for them to hear the alarm.
3. The user is scrolling his/her phone, unaware of the number of whistles already out - People who simply want to be on their smartphone and be oblivious to the cooker whistles in the background, can opt for a simple push notification when the whistles have been counted.
4. The cook has some urgent work, and wants to delegate the task of counting the whistles to someone at home - This is a very common scenario, and in such case one can leave their smartphone near the kitchen and choose to notify someone else via SMS or email when the cooker whistles have been counted.

In all these cases, the person who is notified has to simply switch off the cooker, eliminating the need of relocating to somewhere near the kitchen and disrupting their work.

based on these insights, I started designing the interfaces. Given the utilitarian purpose of the app, I wanted to make it as simple as possible for anyone to jump in and start using it. Here are some key factors which drove the interface design process

1. Clean, high-contrast UI for easy readability by elder generations who want to delegate the task of counting whistles.
2. Declarative UI structure for simple, no-nonsense app usage.
3. Forgiving and flexible UX

🔖 branding

branding assets for Whistl App

I used black as my primary colour, tints of black to represent hierarchy and San Francisco as my primary typeface as people are usually more comfortable with the system font.

🎨 app design

Low Fidelity prototypes for Whistl App

The lo-fi prototypes show a rough schematic of how the app functions. The information flow on the home screen is conversational, to make it very easy for people to jump in and start using the app. The notification methods are present on the home page itself, and let you set your preferences when you first choose them. The whistle detection screen has a real-time view microphone output view which shows whether the model detected a whistle, and also provides a timeline of whistle detections. The timeline is helpful to get a temporal estimate of when was the last whistle detected. The app also lets you customize the number of whistles you want to count, on the fly. This is helpful in case you change your mind or feel the food should cook for longer.

🔧 testing

In my first prototype testing, I got the following feedback from users.

1. Some people said they preferred getting a phone call when all whistles have been counted.
2. The ability to notify someone is useful, but it will be better if they get notified through an alarm of some sort, as they miss out on a SMS or email notifications.

Phone calls and alarms are naturally something we prefer as they are relatively more intrusive and better at grabbing people's attention. But notifying someone else through an alarm would require a client app installed on the recipient who is to be notified, which complicates the app's functionality. Instead, I added an option where the user can notify someone through an automated phone call, which is simpler to use and implement. After making these changes, I designed the final UI.

Home Screen and Main App Screen

Functional flow for Whistl

Notification preference popups on the homescreen

Various notification customization features

I am currently testing the iOS app and fixing a few bugs before launch. As the Indian smartphone market is heavily dominated by Android users, I have planned on rewriting the app in Flutter, a cross-platform development framework, to reach a wider market and have a bigger impact.

It might seem like this study might have been an overkill for the problem in hand, but through this laborious process, I was able to uncover various kitchen practices in India, debunk myths about how pressure cookers should be used and create a robust and accurate app for helping people.

We have problems everywhere, it's just a matter of noticing them and solving them.

Footnotes

1. Sengupta, Subhodh Chandra; Basu, Anjali, eds. (January 2002). সরোজনলিনী দত্ত [Saroj Nalini Dutt]. Samsad Bangali Charitabhidhan (Bibliographical Dictionary) (in Bengali). Volume 1 (4th ed.). Kolkata: Shishu Sahitya Samsad. p. 61 ↩