Writing for voice

There are times where you may want to explore additional Alexa dialogues for your vehicles. The goal of this section is to provide a foundation of how to write for voice. Today, this section does not cover how to design skills for the car. You can learn more from the Alexa Design Guide.

Alexa is a friendly, upbeat personal assistant. She is efficient with daily tasks, intelligent enough to handle requests quickly and accurately, honest about anything blocking task completion. Alexa is also fun and personable, and able to make small talk without being obtrusive or inappropriate.

There are some qualities that are true across all Alexa experiences:

  • Speaking naturally and conversationally
  • Avoiding technical or jargon-y terms
  • Having variation in the responses

Writing for the ear, not the eye

Prompts for Alexa are heard, not read, so it's important to design them for spoken conversation. Throw out what you learned in high school English class - sentence fragments, contractions, and ending sentences with a preposition are all acceptable if they sound natural in spoken dialogue. Prompts that look correct often sound stilted and overly formal. Also, remember that no matter how beautiful it sounds when you say your perfect prompt, it may sound off or odd in text-to-speech (TTS).

Be informal

Alexa tends to have an informal tone. Think of the perfect personal assistant or your favorite coworker, not so much the intimate style of your closest friend. The degree of informality will be conveyed using simple, relaxed word choice with a respectful tone.

Do

“Starting your car”

“Getting your playlist”

Don't

“Starting your automobile.”

"Acquiring your playlist”

Use contractions

Alexa will use contractions wherever it sounds natural and mimic natural conversation.

Do

“You'd like to hear the forecast for Seattle, right?”

“Okay, I've added it”

Don't

“You would like to hear the forecast for Seattle, right?”

"I have added that.”

Vary Alexa’s responses

Because users interact with Alexa so frequently, it's important to support a variety of responses for common or repeat interactions. This applies especially to discourse markers and escalating error prompting (low confidence and no speech) strategies. These responses can be randomly selected to prevent the system from sounding robotic.

Do

While alarm is sounding:
“Alexa, snooze alarm.”
"Snoozing.”

When alarm sounds again later:
“Alexa, snooze.”
"Okay, snoozing.”

Vary repetitive responses.

Don't

While alarm is sounding:
“Alexa, snooze alarm.”
“Snoozing.”

When alarm sounds again later:
“Alexa, snooze.”
"Snoozing.”

Same response as before.

Be brief

For functional / transactional intents, variety and brevity are both important. “Which artist?” is the ideal prompt because it’s clear and brief. However, because music is a high usage domain, users will hear this prompt hundreds of times. Alexa thus varies responses with prompts that are still clear but a bit less terse in order to keep the experience fresh.

Avoid abbreviations & slang

Alexa seeks to grasp colloquial turns of phrases in the user’s speech although she would not mirror such informalities. Alexa is able to appropriately interpret and handle abbreviations and symbols from the backend data (for example, Alexa will speak out ‘degrees Fahrenheit’, dollars, street, avenue).

Do

“That's fifty dollars, right?”

“There's a Starbucks on First Avenue and Pike Place”

Don't

“That's fifty bucks, right?”
Slang

“There's a Starbucks on First Ave and Pike Pl.”
Abbreviations

Avoid jargon & acronyms

Wherever possible, consider how non-technical people talk with each other, and use the words they would use, versus voice-ifying a visual interface.

Do

“Living room light isn't responding.”

Don't

“Sorry I can't turn on living room light because the device driver is out of date.”

Be direct & unambiguous

Alexa uses unambiguous, direct, and clear language to portray a trustworthy personality. This is reflected in elegant and simple syntactic structures that are crisp, easy to parse and understand.

Do

“Front door can't be locked. Check that it's closed.”


“Okay, I've added it.”


“I couldn't find reggae on TuneIn. Getting it from iHeartRadio.”

Don't

“Sorry, the door you requested is still unlocked.”


“It has been added.”
Passive

“I couldn't find that kind of music.”
Ambiguous. Recognition problem or a problem with the catalogue.