A new audio-based social network with a seamless, endless feed.

Jump to…

• What is Odd.io

• Researching Users’ Relationship with Audio Media

• Prototyping Odd.io

Methodology:

A/B testing, Storyboarding, Interviews, Surveys, Speed dating, Experience prototyping

Problem Space

Odd.io fulfills unmet user needs in the current social media ecosystem. Screen fatigue is an increasing problem as more people are spending more time on screens. The volume of content posted on social media is constantly increasing, but users have a desire to reduce the portion of their day they spend looking at a screen.

Storyboards and speed dating

The team conducted two rounds of storyboarding to test and verify our initial hypotheses on how people wanted to consume audio based social media content. The interactions that we looked into included commenting, audio editing (such as voice filters), voice controls, gesture and visual controls, and more.

Each storyboard was built around a “how might we?” statement. We speed dated the storyboards wherein the facilitator showed the individual participants each storyboard one at a time. As the participants viewed each storyboard, they were asked to speak about their initial reactions to the story, about how they felt about doing the actions in the storyboard themselves.

Round 1 Results

Positive Reactions: There were extremely positive responses to the idea of humorous, voice distorting audio filters. People were also interested in to listening to audio threads as a way to interact with their friends on a social media platform. This idea was predicated on the idea that they would be listening to their friends, not to a transcribed audio read by a CUI (such as Alexa or Google Home). People seemed receptive of, and comfortable with, the idea that some sort of visual would be involved in the platform. The final idea that was positively received by the participants surrounded the ability to share clips of audio/podcast content with friends.

Negative Reactions: We also found three areas to which people had strong negative reactions. People did not like the idea of using clapping or gestures to control an audio based social media site. To many, these interactions presented far too large an opportunity for error. There were also concerns about decreased functionality without a visual interface. Without a visual, people worried that they would make mistakes, and participants expressed a large fear that they would accidentally upload something to the social media platform.

Round 2 Results:

Positive Reactions: We received positive responses from all of the participants to the idea of funny audio based sound filters. This is an idea that also received positive feedback in the first round. In the second round, we presented ideas that were received well the first time, as well as new ideas. One of the new ideas that was received very positively, is the idea of having integrated visual aids that convert the audio to text. Participants felt that this would increase accessibility and usability during situations where listening to audio is not feasible.

Negative Reactions: The idea that received the largest amount of negative feedback in this round was about environmental sound needs, as the participants largely felt that need was already covered by existing options.  

Survey

Thirty-seven people answered the survey of sixteen questions with a mix of quantitative questions (how many people choose X) and qualitative questions (open-ended responses).

Key takeaways:

1. 97.3% of the participants indicated that they usually multitask when playing audio. Suggesting that people listen to audio as a secondary or tertiary task used to complement a primary task.

2. People avoid opening audio content in a public setting when they are not wearing headphones. Participants don’t open audio and video on social media in public because they don’t want to disturb others if they don’t have their headphones or earphones in. 

3. Audio is most likely to be consumed while people are in commute, cooking, or doing work.

4. People are interested in listening to stand-up comedy, talk shows, and event commentary on an audio social media platform. 

The survey informed us people don’t have a strong preference between using audio social media hands-free or with GUI. Short answer responses fueled the assumption that these preferences are informed by infrequent usage and a general distrust of conversational agents.

Pretotype

Based on the results of our previous methods, the team developed a “pretotype” to gauge how users would actually feel about specific elements of our idea. We wanted to understand how people would react to threads of their friend’s audio. This was something that people expressed interest in, but we wanted to understand how an actual thread would affect people. People were interested in the idea, but without any current example of the idea, people had no experience with which to consider this idea. 

After listening to the string of audio clips, each participant was asked questions about their listening experience such as: “What did you think of these clips? Would you listen to something like this? How would you like to interact with the audio?”.

We told each participant they were about to listen to an audio-only social media app, in which their friends were talking about their day. 

Parallel prototype testing

We designed the prototyping phase to analyze CUI interactions on social media platforms, and understand how users listen to posts, comment on posts, create posts, and adjust their feed. Parallel prototyping tested two potential formats for audio based social media platforms. One prototype was a conversational user interface (CUI) and one mixed conversational and graphic user interfaces (GUI).

Both prototypes tested commenting and categorization features, and used “Wizard of Oz” to simulate working CUIs. We conducted eight sessions, four participants were asked to complete tasks for the CUI, and the other four were asked to complete the same tasks using the mixed CUI and GUI.

Overview

As illustrated in the research sections above, we started with a simple audio-only prototype, a demo version of what one’s feed might sound like. We used feedback on this prototype to better understand the overall viability of the concept, as well as what situations and settings users see themselves using Odd.io in. Users highlighted active situations (driving, cooking, working, cleaning, etc.) as key times to use Odd.io. This pointed us in the direction of a hands-free or audio-based interaction system that would let users perform some familiar social media tasks such as liking, saving, and responding to posts.

In our next iterations, we tested specific interactions in audio and visual format to understand which features were most desired, which interactions required visual interfaces and feedback, and which could be accomplished via a conversational user interface. Ultimately, we found that users were comfortable accomplishing basic tasks (passive listening, liking, following, and feed navigation) in a pure audio CUI, while a visual interface could be a home for less frequent or more nuanced interactions like editing one’s social graph, adjusting content preferences or creating richer posts.

This led Odd.io to be a mix of visual and conversational interface, but include a hands-free, CUI only, mode for listening, browsing, and commenting. Additional features added to the final prototype based on previous testing include:

1. Adjusting GUI and VUI elements to provide better feedback and feedforward

2. Providing audio-to-visual transcription

3. Commenting in real-time audio on audio clips, allowing for a targeted experience

4. Providing an array of freely available audio reactions for an enhanced experience, similar in purpose to emojis and gifs

5. Creating audio snippet sharing options, so users can send and receive relevant pieces of clips within their social network 

Odd.io’s latest iteration

How does odd.io work?

Back to Top