Feature to set volume level for an announcement

[danielbrunt57]

I've worked out code to allow you to set a volume level for an announcement after which the previous volume level is automatically restored. Here is the revised notify.py for you to test: notify.zip

To pass the temporary volume level, you code it like this:

action: notify.alexa_media
data:
  message: announcement
  data:
    type: announce
    volume: 0.4
  target:
    - media_player.office_echo_dot_left
    - media_player.kitchen_echo_dot

[popy2k14]

@danielbrunt57 first, happy new year!

Thx for your implemetation.
I have tested it. It set's the volume before sending the announcement.
This lead's to shortly jumping of the volume of the current playing media (tunein, ...).

My script from here: #2766 (comment)
Is doing this some trickery to prevent this.
But it's handling all announcements synchronous when sending to more targets.

I have now made a version which respects various delays and also prevents the volume jumping of the music.
The version also is running announcements/tts in paralell (async), which is a nice addition if you have multiple echos (one i every room).
Had to change (remove) the queue delay to get precise timing.
I am not so deep in the api limitations and hope this makes no issues.
Do you know this?

Also tried it with different sentence lenght's (in german) and optimized the end delay.
Can you please test this version with multiple echo devices, like this:

data:
  target:
    - media_player.wohnzimmer
    - media_player.kuche
    - media_player.badezimmer
  message: "Waschmaschine ist fertig, bitte ausräumen!"
  data:
    type: announce
    volume: 0.4    
enabled: true
action: notify.alexa_media

Before, please manually start an tunein station with low volume like 20 (example like in the bathroom in the morning).

Download: notify.zip

One issue i noticed (and is currently present in my version) is: on tts the "targets" attribute does not work correctly with mutiple devices.
So just the first device in the targets list get's the tts command.
Any idea's/hint's how we can extend the tts functionality to take more target's (like announce does)?

[danielbrunt57]

One issue i noticed (and is currently present in my version) is: on tts the "targets" attribute does not work correctly with mutiple devices.
So just the first device in the targets list get's the tts command.
Any idea's/hint's how we can extend the tts functionality to take more target's (like announce does)?

This is the first thing that I tested in my version and multiple targets works:

action: notify.alexa_media
data:
  message: this is a message
  target:
    - media_player.office_echo_dot_left
    - media_player.living_room_echo_dot_left
    - media_player.kitchen_echo_dot
  data:
    type: tts

I have not had a chance to look at your version yet...

[popy2k14]

Yeah, I know.
Your version was Handling multiple tts devices in an for loop and the async tts call just received one target at once. The announcement call has a break and adds all targets at once to the async announce call.

In my version I have used the single call approach for both (tts and announcements) so both can benefit from the "loud music fixes".

As stated, the async tts call does currently not handle multiple targets, that is a thing we have to address (look at it)

Looking forward to your announcement tests (preferred with multiple devices).

[danielbrunt57]

Had to change (remove) the queue delay to get precise timing.
I am not so deep in the api limitations and hope this makes no issues.
Do you know this?

The queue_delay was designed (I think) to reduce the number of API calls to help avoid error 429.
Hopefully your change does not induce that error!

[popy2k14]

ok, thx i have added it back but because timing is key here, sometimes the music volume now pop's up again.

[danielbrunt57]

I just had a look at your version and I see the problem re: tts.
tts has to process each individual target separately sending them the tts message
announce just uses the fist echo entity as the "controller" and passes it the list of entities to announce to, hence the break at the end of the for loop.

If we want tts to support volume, it has to be designed similarly to announce but without a break (like the original code), or, maybe you can adapt the loop into your version?

            for alexa in account_dict["entities"]["media_player"].values():
                if data_type == "tts":
                    targets = self.convert(
                        entities, type_="entities", filter_matches=True
                    )
                    # _LOGGER.debug("TTS entities: %s", targets)
                    if alexa in targets and alexa.available:
                        _LOGGER.debug("TTS by %s : %s", alexa, message)
                        tasks.append(
                            alexa.async_send_tts(
                                message,
                                queue_delay=self.hass.data[DATA_ALEXAMEDIA]["accounts"][
                                    account
                                ]["options"].get(CONF_QUEUE_DELAY, DEFAULT_QUEUE_DELAY),
                            )
                        )

for alexa in account_dict["entities"]["media_player"].values() loops through all media_player devices but only processes those in the list of entities we provide (if alexa in targets) when making the call. This results is separate alexa.async_send_tts( commands for each media_player we specify and each individual command looks ike: media_player.office_echo_dot_let.async_send_tts(.

[popy2k14]

I have fixed the tts issue just by adding the loop a second time.

Here is my current version: notify.zip

Sadly the restoring does'nt work correctly sometimes, so music stays on the speak volume (especially when using multiple targets).

I think there are multiple issues with this:

• to high api demand (sometimes the previous volume get's not updates in time)
• queue delay
• if you specify muiltiple targets, volume and as a result previous_volume get's not updated in time

My initial script is announcing/tts'ing all alexas synchrounous (one after one) and not parallel, like this approach.
My skills in python and alexapy sadly are low, so i will not try to solve those issues in the meantime.

Maybe you could take my latest version and get it to work stable?

P.S.: It's so sad amazon was just to lazy to add an "volume" attribute to the announce/tts :-(

[danielbrunt57]

The applestrudel package using alexa-remote in Node-RED has this node:

and logging shows a single POST with set new volume, the TTS to speak, and set the old volume:

2 Jan 16:12:42 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote: Sending Request with {"host":"na-api-alexa.amazon.com","path":"/api/behaviors/preview","method":"POST","timeout":10000,"headers":{"Accept-Language":"en-CA"}} and data={"behaviorId":"PREVIEW","sequenceJson":"{"@type":"com.amazon.alexa.behaviors.model.Sequence","startNode":{"@type":"com.amazon.alexa.behaviors.model.SerialNode","nodesToExecute":[{"@type":"com.amazon.alexa.behaviors.model.ParallelNode","nodesToExecute":[{"@type":"com.amazon.alexa.behaviors.model.OpaquePayloadOperationNode","type":"Alexa.DeviceControls.Volume","operationPayload":{"deviceType":"A2H4LV5GIZ1JFT","deviceSerialNumber":"G091ET10051402M6","locale":"en-CA","customerId":"A1UZ93DGNU3F5L","value":70}}],"name":null},{"@type":"com.amazon.alexa.behaviors.model.OpaquePayloadOperationNode","type":"Alexa.Speak","operationPayload":{"deviceType":"A2H4LV5GIZ1JFT","deviceSerialNumber":"G091ET10051402M6","locale":"en-CA","customerId":"A1UZ93DGNU3F5L","textToSpeak":"Hello"}},{"@type":"com.amazon.alexa.behaviors.model.ParallelNode","nodesToExecute":[{"@type":"com.amazon.alexa.behaviors.model.OpaquePayloadOperationNode","type":"Alexa.DeviceControls.Volume","operationPayload":{"deviceType":"A2H4LV5GIZ1JFT","deviceSerialNumber":"G091ET10051402M6","locale":"en-CA","customerId":"A1UZ93DGNU3F5L","value":20}}],"name":null}],"name":null}}","status":"ENABLED"}

2 Jan 16:12:42 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote: Response: No body (code=200)


2 Jan 16:12:43 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote HTTP2-PUSH: Command PUSH_VOLUME_CHANGE: {
    "dopplerId": {
        "deviceSerialNumber": "G091ET10051402M6",
        "deviceType": "A2H4LV5GIZ1JFT"
    },
    "volumeSetting": 70,
    "isMuted": false,
    "destinationUserId": "A1UZ93DGNU3F5L"
}


2 Jan 16:12:43 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote HTTP2-PUSH: Command PUSH_VOLUME_CHANGE: {
    "dopplerId": {
        "deviceSerialNumber": "G091ET10051409C4",
        "deviceType": "A2H4LV5GIZ1JFT"
    },
    "volumeSetting": 70,
    "isMuted": false,
    "destinationUserId": "A1UZ93DGNU3F5L"
}


2 Jan 16:12:44 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote HTTP2-PUSH: Command PUSH_VOLUME_CHANGE: {
    "dopplerId": {
        "deviceSerialNumber": "G091ET10051402M6",
        "deviceType": "A2H4LV5GIZ1JFT"
    },
    "volumeSetting": 20,
    "isMuted": false,
    "destinationUserId": "A1UZ93DGNU3F5L"
}


2 Jan 16:12:44 - [debug] [alexa-remote-account:14f19d12849e6414] Alexa-Remote HTTP2-PUSH: Command PUSH_VOLUME_CHANGE: {
    "dopplerId": {
        "deviceSerialNumber": "G091ET10051409C4",
        "deviceType": "A2H4LV5GIZ1JFT"
    },
    "volumeSetting": 20,
    "isMuted": false,
    "destinationUserId": "A1UZ93DGNU3F5L"
}

• There are two volume updates since echo_dot_left and echo_dot_right are a speaker pair

** OpenHAB (Java-basesd) also has a method to speak/announce at a different volume level.

[popy2k14]

When i understand the code right, from here: https://github.com/bbindreiter/node-red-contrib-alexa-remote2-applestrudel/blob/2cfc43d68d0721f1767aec7ca8a14a32b4dab406/nodes/alexa-remote-routine.js#L150
it seems that the routine just set's the volume before the speak and not reset's it.

[danielbrunt57]

There are two options in Speak At Volume: Set and Add. I presume Set sets it and leaves it there. I chose Add without knowing what it would do as the name did not mean anything to me.

Now that I see the code for add, there are two "volume" commands... one before "speak" and one after:

			if(node.payload.mode === 'add') {
							return await nativizeNode({
								type: 'node',
								payload: {
									type: 'serial',
									children: [
										{
											type: 'node',
											payload: {
												type: 'parallel',
												children: devicesWithVolume.map(device => ({
													type: 'volume',
													payload: {
														value: tools.clamp(deviceToVolume.get(device) + node.payload.volume, 0, 100),
														device: device,
													}
												}))
											}
										},
										{
											type: 'speak',
											payload: {
												type: node.payload.type,
												text: node.payload.text,
												devices: devices,
											}
										},
										{
											type: 'node',
											payload: {
												type: 'parallel',
												children: devicesWithVolume.map(device => ({
													type: 'volume',
													payload: {
														value: deviceToVolume.get(device),
														device: device,
													}
												}))
											}
										}
									]
								}
							});
						}

[danielbrunt57]

I've studied the code further and now I see the difference between add and set!
set just sets the volume to the volume specified.
and
add uses value: tools.clamp(deviceToVolume.get(device) **+** node.payload.volume, 0, 100)

add adds the volume level specified to the devices current volume level.
So if the devices volume_levels are currently:

device_1: volume_level: 20
device_2: volume_level: 35
device_3: volume_level: 30

the announcement using add volume_level: 30 occurs at:

device_1: volume_level: 50
device_2: volume_level: 65
device_3: volume_level: 60

[popy2k14]

Nice find!

But will the volume change handled by the device itself (and is smoother and no jumping in music/stream volume)?

[danielbrunt57]

Nope. If music is playing, the volume is set/increased, the TTS/announcement occurs all the while music continues playing at the new volume level while the text is spoken, then volume level is restored. The only solution I found for this in NR is:

That would have to modified to first test if music is being played, pause it, then resume it.

I've no idea if OpenHAB handles things differently.

It's all quickly getting very complicated...

[popy2k14]

That is a good idea to check if anything is playing, if yes, pause it and resume it.

Maybe we should try tob implement it this way in notify.py?