Some skill interactions last longer than a simple utterance or even a dialog. For example, let's say that a skill has been created to act as an interactive guide to a National Park. The user may want to interact with that session for several hours while they explore the National Park, asking for information about the areas of the park they are at.

For extended sessions with a skill, it should be possible (given user permission or a specific invocation phrase) to keep a session open without the skill leaving the microphone open and without the user speaking the invocation name on every interaction. It should stay open until the user explicitly closes the skill or some extended timeout period (say, 24 hours).

E.g., if they are in Yellowstone, they may ask for information about Old Faithful, and then as they walk further down the trail ask for information about Grand Geyser, and then a little later ask about Morning Glory Pool. After leaving the Upper Geyser Basin, they may arrive at Mammoth Hot Springs an hour later and ask about some of the features there. In this case, they could start their experience by saying "Alexa, open Yellowstone Guide for an extended session" and then until they explicitly say "quit" (or some other utterance triggering the stop intent) they can ask questions without saying "Alexa" every time.

Alexa can ask after a period of time with no response, "Do you need more time?"

I think this will be possible only when a skill can install custom temporary wake words, because longer listening time (>8 sec) is not acceptable for privacy reasons. However, I do hope that a solution like that given by Marius Narkus above is implemented, a special type of skill that allows longer listening times if the user has agreed beforehand, and where frequent feedback about the listening status is given.

My motivation is a game skill which is made unplayable by the discussed restriction in listening time, or on the other hand by lack of customizable skill-specific wake words.

My mother-in-law had a stroke a few years ago and can only blink. She has a special keyboard that allows her to select letters, write, and turn the text to audio. Alexa can recognize the generated audio. The problem is that my mother in law takes a long time to write, so i need to make Alexa able to listen for longer.

Please implement this feature. It would mean the world for people with disabilities. I live in Brazil, so please make this available here. Thank you so much.

I think one thing that would help some of the concerns expressed would be an adjustable time before Alexa assumes a command is complete, so that people who speak more slowly can more easily interact. Instead of having to say "wait a sec" Alexa would merely wait one additional second before responding to a request.

I guess the problem why amazon has not already implemented this feature is that they are afraid someone could create a skill which will ask the user to say "wait" and then spies out the room.
To encourage Amazon to implement this I would suggest 2 security features which could be implemented with this:
1) Alexa could play a beep every 3 seconds to remind the user that Alexa is waiting for a response
2) Every skill using the wait feature could have a mandatory user confirmation: "Yes, I have read and understood what that feature does in this skill"

Taking this to account could make the feature even cooler if it would not be necessary to say "wait". Alexa could just wait longer than 8 seconds.

Allow for the user to talk for a longer period of time, and allow brief pauses.

I know these two issues have been previously raised separately. My goal is to provide another use case covering both and remind Alexa team of the relevance of these feature requests.

---
I've recently published my skill called "Run My Lines", allowing actors to learn their lines using Alexa (https://runlines.me/). I've promoted it on several internet forum visited by actors, including Reddit and Mandy (see links below), and received very positive feedback.

However I am also aware of some critical limitations of Alexa around user input, which prevent it from being fully useful. This includes:
- allowing users to have a pause between sentences (Alexa thinks it's over after a very brief pause),
- allowing users to speak for a longer amount of time (I believe there's a limit of 8 seconds or so?)

Due to current limitations the skill is unfortunately limited in its practicality, as most lines will be longer than a few seconds and contain several sentences (hence short pauses).

- Mandy: https://www.mandy.com/forum/topic/106957/running-lines-with-alexa
- Reddit: https://www.reddit.com/r/acting/comments/dxrkdu/running_lines_with_alexa/

I want to use this for slicing and dicing data with Python! So if I want to get some data and plot it, followed by other actions, I want to be able to send voice commands with some pause in between. So I want Alexa to be on and focussed without having to say "Alexa, blah blah..." each time.At least till I tell it to switch off :D
The follow on mode of just 15 seconds does not work for this, I could be thinking for a min or more. So it should be configurable - maybe upto 5 mins?

Excellent idea - it would be quite useful when formulating questions.

I'm not a native English speaker and it takes me longer to think about the command I'm going to say. As such, I often find the follow up mode unusable for me because Alexa stops listening sooner than I manage to formulate the next command in my head. It's also often a bit unclear when does Alexa start listening (in Follow Up) and when does it stop so I easily end up in ridiculous command-repeating situations.

Making this listening time configurable would allow me to extend this duration while keeping it the same for quicker speakers.

Additionally, I'd like the skill to be able to dictate this. I currently have a problem where the app version of Alexa resolves the listen() method within 1 second (instead of the 8 second wait for a hardware device). The user is not able to make a reasoned reply in this time.

Good for cognitive accessibility

Yes,
Even if you add an extra session end (true, false or pause)
and have to add resume speech parameters.
It could even end the conversation completely, and if the user's next request to that machine matches the resume parameter, pick up where you left off.

And just talk slower too. Adjustable speed would be of much help to older users, those hard of hearing, kids, and just tired people.

This is a very important element of a "real", practical, and respectful user interface. Imagine a computer screen offering an option to the user, then taking it away if they don't decide within 8 seconds!

Or, if they take 2 tries correcting what "the computer misunderstood" (from the users point of view), then she just hangs up on you, instead of giving you unlimited opportunities.

This is REALLY important!!

I think a pause handler is a good idea or at least a way to increase the timeout like we can on the Lambda function.

YES

Really it would be a great thing!

I'm currently developing a skill implying long conversations with Alexa and some réflexions.

Being able to pause the skill because we are interrupted, or because we need to think would be great.

I would think to two new special handlers: PauseHandler (to define behavior just before going to pause), and RestoreHandler (behavior when relaunching a paused skill, to remind the user the status or the question for instance before continuing).

Excellent idea. With some skills I've messed around with in and out of development, I've needed a couple extra seconds to think.

Simple example: The well known Jeopardy skill and other trivia skills. It gets a bit distracting to have Alexa's reprompt/timeout happen when I'm trying to think of an answer, so being able to say "hey Alexa wait a sec" would take care of that.

Or sometimes, people might open a skill (without any intent initially) and need some time to think about why exactly they opened it. In a perfect world, people might think about it before opening every single time without fail, but sometimes, at least in my experience, people will open a skill because they know they need something from it, but haven't exactly worded it in their head what they want. I've had a good laugh a couple times hearing someone go "Uh........" after loading it and having alexa interrupt their thought process because of the timeout. Haha.