Glossary: Alexa Narrator
Published in
3 min readMay 20, 2018
- Context
Choosing the right narrator or moderator for a voice game - Definition
‘Alexa Narrator’ is a possible choice about the storytelling of a voice game, in which the voice of Amazon Alexa is used for leading the player through the game situations.
It can also be a (sometimes involuntary) choice about the composition of the game, especially if a first or second person character point of view is used to narrate the story or moderate the game.
Typically, users don’t differentiate the the Alexa voice from the Alexa personality that Amazon builds through their native responses and first-party Skills. This can be either an advantage or a disadvantage, as detailed below. - Advantages
To start with the obvious: Using Alexa’s voice has no costs even at arbitrarily heavy usage, and works out of the box for Alexa Skills.
Text-to-speech (TTS) with Alexa’s voice is very mature, and sounds both less mechanical and more pleasing than most other TTS voices.
The strong bond between Alexa’s voice and personality can be an advantage, in so far as it gives a Skill with Alexa’s voice a small trust and rapport bonus. It also opens up the possibility of the Skill revealing and exploring a new aspect of Alexa, as long as it is consistent with her character. - Challenges
Alexa speaks individual sentences sufficiently expressive, but won’t adapt one sentence’s prosody according to the text’s structure, i.e. the prosody of neighboring sentences — Which ultimately makes her sound increasingly monotonous with the length of the text. This severely limits the amount of text a game can use without potentially losing engagement (that being said, you can make longer texts enjoyable by manually fine-tuning the SSML, inserting pauses or audio effects, and using direct speech / dialogue).
The strong tie between the Alexa voice and personality might also render her unfit for certain atmospheres or roles. For example, she might be too light-hearted for a adrenalin-rich themes like dungeon exploration or chase scenarios, and she might be too kind and mild-tempered for stereotypically masculine roles like a soldier or a fire fighter.
A minor technical limitation is that with the Alexa voice, you can use only a subset the SSML tags available for other TTS services like AWS Polly, e.g. you can’t control her timbre (i.e. vocal tract length) or most of her breathing.
Another issue to be aware of is that Alexa might not pronounce everything as intended — Typically not because of glitches in the TTS engine (even though those also happen), but because of the ambiguity of the input (e.g. is ‘9.05.’ a date, a time or a decimal number followed by a period?). This necessitates some degree of quality control and fine-tuning.
Just for the sake of completeness: Alexa’s voice can only be used in Alexa Skills (unless you’re doing something fun, in this case, please let me know!). - Related terms
Alternatives: Custom Text-To-Speech Narrator, Voice Talent Narrator - Examples of effective use
(Note: These voice games might or might not be great, but they use this feature very well)
Mein Auftrag
The Magic Door
Kids Court - Examples of sub-optimal use
(Note: These are great voice games, but they might not have reached their full engagement potential in regard to this feature)
ALIEN Offworld Colony Simulator
Akinator
