Right now, in the test tab for a given skill in the developer console, there's an option to provide some SSML text and hear how Alexa would convert that to speech. This utility uses no contextual information about the skill itself, and there's really no reason at all that it needs to be stuck behind a login gate, other than for the purpose of possibly thwarting abuse of said functionality.

As a developer, I would like the ability to provide this same functionality to my users on a public website I run or to my collaborators in a private tool of my own.

Preferably this would exist in the form of an embeddable widget that I could just drop into my site, but an API that would let me do the same thing would be acceptable.

Some sort of rate-limiting mechanism to prevent abuse of the API would be acceptable (and even expected). The ability to choose Alexa's voice by locale (or even a potential integration with Polly voices) would be a nice bonus feature.

To provide a specific use case - I'm looking at building a skill that can consume a small amount of user-generated content, specifically with regard to online handles. I'd like users to be able to test how Alexa will pronounce their IPA-written names before saving and submitting on a website.

I was surprised to find that there is no official Docker image for development with the ASK CLI. The "amazon" user on Docker Hub contains only a couple images for working specifically with ECS and Elastic Beanstalk, and there does not appear to be any official "alexa" user. Upon further investigation, I found the aws-codebuild-docker-images repo in the "aws" GitHub organization, but unfortunately it does not have any sample Dockerfiles that install the ASK CLI either. There are similar issues open at the AWS CLI repo, because there is no official Amazon Docker image for the AWS CLI either, so I imagine that whatever aws-cli image gets created could serve as the base for an ask-cli image.

TL; DR
I believe there should be an official ask-cli Docker image maintained by Amazon and pushed to Docker Hub/Store. Ideally, it should support all the major modern Linux distros, including Alpine Linux. One should be able to run the following command, and have everything just work:

docker run -it alexa/ask-cli

Currently if a user streams audio via the skill there is no friendly way to get back to the skill without the user saying an invocation command.

I'd propose a way to allow a skill to open a new session, allowing speechOutput after the PlaybackNearlyFinished or PlaybackFinished events occur in the AudioPlayer interface.

There are a couple use-cases for this:

1) A podcast skill where it would be beneficially to ask the user if they'd like to listen to the next episode.
2) Give some information about the next episode before it begins to play
3) Provide a purchase opportunity after an episode completes but before the next begins.
4) Allow the user to interact with some other aspect of the skill other than audio (e.g. Tell me about this podcast, or I want to donate to this organization).

"STOP" and "PAUSE" have two very distinct meanings:

"STOP": cause (an action, process, or event) to come to an end.
"PAUSE": a temporary stop in action or speech.
Different words, different meanings.

Unfortunately, AudioPlayer confuses the two, and even the documentation warns that

"when your skill is not in an active session but is playing audio, or was the skill most recently playing audio, utterances such as 'stop' send your skill an AMAZON.PauseIntent instead of an AMAZON.StopIntent."

This creates a poor user experience, and it doesn't take long to find at least 30 questions, requests relating to this.

For example, when a user is playing audio and they want to stop, they mean stop. As in, cease playing. However, when they "pause", they usually want to do something. In which case, the skill could wait for a command. Then the user could say "how long is this track" or "skip forwards 30 seconds".

This would at least provide a partial fix to the request below, although implement the idea below would provide a far better solution in all cases.

https://alexa.uservoice.com/forums/906892-alexa-skills-developer-voice-and-vote/suggestions/33406042-remove-requirement-for-invocation-name-when-an-aud

Adding the ability to specify a font color in addition to the existing font size would be very helpful. I was developing a test skill for my company an on the prompt I used the company logo as a background image. Unfortunately the logo is on a field of white and the text was invisible since the display font is white with no option to change it. I naively assumed I could put a rich text option <font color="black"> into the text but discovered there is no such option. Just the size. Color setting would be very useful for Cards, List Templates and Body Templates.

There are multiple Asks from other Users over a number years, yet Amazon isn’t acting or reacting.

Alexa Context Awareness (Room Awareness) for any device placed in a group + Echo device, to control using natural utterance as long as it is in a group using the object name. (same as lights/switches).

Alexa supports Lights, Switches, Thermostats - today. However myself and many others would like to Control the devices we choose to insert into a group with natural utterances, regardless if they are a Blind/Shade, Awning, Fan, drape etc but use the objects name vs having to utter Alexa Close Blind “XXX1”, then in another room Alexa Close Blind “XXXX2” etc. Amazon should support this to place any device into a group and as long as an Echo device is in the group it has (context / room awareness), using natural utterances to control "Alexa close/open blinds" or “Alexa Raise/Lower Blinds”; “Alexa Turn Fan on/off” just as it works with lights: "Alexa turn lights on/off". You can’t solve this in a routine, its clunky, they are not being adopted by my family as they tell me it’s too complex and not intuitive, plus you need to remember a routine for every room.

If you could support Context Awareness / Room Awareness for Devices placed into a Group in which the Echo device resides, then use natural utterance to control the device(s), without having to use unnecessary naming conventions & utterances. This seems like such an easy fix / implementation and long overdue. Doing this would also encourage more people to purchase more devices to be controlled this way and would also increase Echo insertion & adoption into households.

Please, please can we get this implemented