We live in a home where there are constantly distracting noises. Two freeways, train tracks, and BART tracks (city transportation) are nearby. A firehouse across the street. Our neighbors are musicians. We are also close to the bay, and oil tankers going back and forth to the Mobil station are constantly sounding their horns.

Regular white noise generators don't cut it, when it comes to blocking some of this undesired ambient noise. As a result, we tried making different noises for ourselves. We ended up with deeper noises, which help block unwanted noise. We tried a wide variety of public domain clips, including recordings of National Park waterfalls, geysers, jet engines, steam engines, rocket launches, and even volcanoes. We narrowed down our choices to three clips, which are a part of this skill.

As for naming the individual sound options, we went with "Black Noise" because they are lower pitched, deeper, more rumbly and lush than typical white noise tones. Within the skill, we also gave each one a name: "Black Noise" for the sound based on a Swedish waterfall. "Blacker Noise" for the sound based on a Yellowstone National Park steam vent. "Blackest Noise" for a NASA Delta IV rocket launch.

We decided to offer these in an Alexa skill to share with others who could benefit from these particular noises as well. Because these are so deep, they are not for everyone. The great thing is that there are many ambient sounds already available in the Alexa Skills store. What this skill offers is several additional options which we believe are very unique and will hopefully be of value for some Alexa users.

What it does

The skill offers three ambient sounds. The free versions of the sounds consist of 30-second samples and 3-minute long clips. The paid/premium subscription offers continuous looping for overnight sleeping and longer-term use.

The file sizes for continuous looping are pretty large, and are stored on Amazon S3. Because of the S3 costs, we needed to also utilize ISP (subscriptions) to help offset the costs.

APL Cards with images of the sound sources are included. This helps make the skill a bit more interesting. These include public domain images of the waterfall, the Black Growler steam vent in Yellowstone National Park, and NASA Delta IV rockets and diagrams. The cards also reinforce visually the available options, and what is being listened to at the moment. However these are not required when using the skill.

When the skill opens, users can listen to the free 30 second samples, the free 3-minute clips, or learn about the premium, continuous looping option. If the user choose to learn about the continuous looping option, an upsell message and offer to buy occurs. Users can also ask "What can I buy" at any time, as well as "What did I buy?" A free trial is offered. Users can also cancel their subscriptions at any time.

How we built it

Voiceflow was used to build the skill. Garageband and Audacity were used to edit the sounds.

The public domain sound clips were extended using Garageband, in order to have continuous looking for overnight sleeping. We also used Garageband to taper off some the low-end frequencies of the sound clips, and adjust the pitch, to optimize them for use with smart speakers such as Echo Dots and Echo Show, without compromising the deepness of the sounds. Finally, we used Audacity to manage the file size and quality, and ultimately the files have been stored on Amazon S3 for access by our skill.

Challenges we ran into

The first big challenge was editing the sound clips using Garageband. The public domain clips are actually pretty short. Garageband was used to extend the sound clip length, and this was probably the most challenging aspect. Mastering the clips so that they sound good on echo devices was also challenging. We had to assume that most people didn't have the larger Echos, with larger speakers. Our focus was to make sure they provided great sound both from the small Echo Dot as well as on the larger Echo Show.

The second challenge was managing the file size. Initially, they were pretty large in order to support continuous play without gaps (audio dropouts) occurring too frequently. We found sound the quality and file size tradeoff required multiple tests to find the right balance.

The third challenge was adding the ISP features, doing our best to meet all of the recommended best practices, and ensuring all the possible paths flowed properly and in a manner that provides a great user experience.

A fourth challenge was the timing of the cards. When they display relative to when interactions occur took a bit of experimenting.

Accomplishments that we're proud of

Getting ISP functioning. Getting the audio clips to sound good on Echo devices. Getting the timing right when displaying cards.

What we learned

The biggest learning for us was working with the ISP tools, and using them to develop a high quality, engaging experience for users.

What's next for Natural Ambient Sounds: Black Noise

We anticipate adding more ambient sounds based on other sources, which fit the concept of "Black Noise."

Built With

  • garageband
  • voiceflow
Share this project: