navidrome-alexa

README

navidrome-alexa

Stream your music from Navidrome UI to Amazon Echo & Alexa devices.

How it works

Navidrome-Alexa is a combination of a UI widget, REST API and Alexa Skill that allow you to stream your music from Navidrome Web UI to Alexa devices like Amazon Echo. See the below diagram for details how components interact with each other.

Why it works that way

See below for more details on Navidrome and Alexa integrations.

ADR001 Alexa Interactions
ADR002 Navidrome Integration

How it looks like

UI still needs a lot of work both on the look-and-feel side and how it integrates with Navidrome.

Installation

Prerequisites

• Public-web accessible address for navidrome-alexa to accept Amazon Alexa Skill API requests
• Public-web accessible address for Navidrome's /rest/stream endpoint
• Reverse proxy with SSL and rewriting support (Caddy is used in example configuration below)
• Amazon account to access Alexa Developer Console to configure self-hosted Alexa Skill

1. Configure skill

This configures self-hosted Alexa skill that calls /skill endpoint of navidrome-alexa application and provides music streams back to echo devices.

• 1.1. Open Alexa Developer Console and authenticate.
• 1.2. Click "Create New Skill" button.
• 1.3. Enter "navi stream" as skill name
• 1.4. On the "Experience, Model, Hosting Service" screen select:
  • Choose a type of experience: "Music & Audio" (picture)
  • Choose a model: "Custom" (picture)
  • Hosting services: "Provision your own"
  • Click "Next" button
• 1.5. On the "Template" screen select "Start from Scratch" and click "Next" button (picture)
• 1.6. On the "Review" screen click "Create Skill" button, wait till skill is created
• 1.7. Go to "Intents" in left side menu and hit "JSON editor", copy & paste alexa-skill.json and click "Save" (picture)
• 1.8. Go to "Endpoint", select "HTTPS" as "Service Endpoint Type" (picture)
  • Enter public https URL pointing to your navidrome-alexa installation ending with /skill (e.g. https://alexa.yourdomain.com/skill )
  • Select SSL certificate type you use (make sure to select wildcard cert type if using it for subdomains)
  • Click "Save" button
• 1.9. Go to "Interfaces" left side menu, enable "Audio Player" and press "Save" (picture)
• 1.10. Click "Build" in top menu, click "Build skill" button, ensure skill builds successfully
• 1.11. Got to developer console root page and click "Copy Skill ID", you will need it for the next step. (picture)

2. Configure application

This configures navidrome-alexa application that provides /skill endpoint to serve music stream URLs to Alexa devices and /api endpoint to control queuing and playback.

• 2.1. Download a pre-built release of navidrome-alexa or build locally.
• 2.2. Run it passing configuration parameters below:

Command line	Env Var	Default value	Description
amazonDomain	NA_AMAZON_DOMAIN	amazon.com	Base domain to use for Alexa API calls.
amazonCookiePath	NA_AMAZON_USER	cookies.data	Path to a writable file to store auth cookies.
amazonUser	NA_AMAZON_PASSWORD	Empty	Amazon account email with Alexa devices, can be left blank if auth cookies already exist.
amazonPassword	NA_AMAZON_COOKIE_PATH	Empty	Amazon account password, can be left blank if auth cookies already exist.
apiKey	NA_ALEXA_SKILL_ID	Empty	Required. API key to authenticate /client calls. User provided, select arbitrary string to match 4.1
streamDomain	NA_ALEXA_SKILL_NAME	Empty	Required. Navidrome public server domain URL.
alexaSkillId	NA_STREAM_DOMAIN	Empty	Required. Skill id to authenticate calls from Alexa. Has to match copied in 1.11.
alexaSkillName	NA_API_KEY	navi stream	Skill invocation name. Has to match name configured in 1.7. JSON
listenAddress	NA_LISTEN_ADDRESS	:8080	Listen address.
logIncomingRequests	NA_LOG_INCOMING_REQUESTS	false	Log API and Skill requests/responses.
logOutgoingRequests	NA_LOG_OUTGOING_REQUESTS	false	Log outgoing (to Alexa APIs) requests/responses. Will leak sensitive data into logs.
logStructured	NA_LOG_STRUCTURED	false	Structured (JSON) logs output

Minimal configuration via command line example:

  na \
  -amazonUser your@email.com \
  -amazonPassword youramazonpassword \
  -apiKey yourlongenoughandsecureapikey \
  -alexaSkillId amzn1.ask.skill.xxxxx \
  -streamDomain https://navidrome.youdomain.com \ 

Note that Amazon may challenge you with CAPTCHA and this will require logging into mobile app from the same network to clear oauth/openid flow. You can also test Amazon Alexa authentication / generate cookie file with meow command

  meow amazon.com your_amazon_user@email.com your_amazon_password

3. Configure proxy

This configures reverse proxy re-write rule that injects navidrome-alexa UI widget into Navidrome UI. Caddy is used as example configuration. Below examples assume that navidrome and navidrome-alexa are running on their default ports and on the same host as Caddy (localhost).

• 3.1 Add public address for navidrome-alexa in Caddy config, this has to match URL configured in step 1.8.

  alexa.yourdomain.com {
      reverse_proxy localhost:8080
  }
  

• 3.2 Add rewrite rule to inject widget into Navidrome's UI. Navidrome-alexa /proxy endpoint simply concats real navidrome UI js with the widget js. Note that main.6c7b5c7f.js script name is Navidrome release 0.52.0 specific and needs to be updated with each release.

  navi.yourdomain.com {
    reverse_proxy localhost:4533
    handle /app/static/js/main.6c7b5c7f.js {
      rewrite * /proxy?proxied=http://localhost:4533/app/static/js/main.6c7b5c7f.js
      reverse_proxy http://localhost:8080
    }
  }
  

• 3.3. Verify that script now contains widget code with curl -v http://navi.yourdomain.com/app/static/js/main.6c7b5c7f.js

4. Configure widget

If widget was injected successfully when opening Navidrome UI, there will be a new button on the player bar.

• 4.1 Click on the speaker button, click on three dots ... (picture)
• 4.2 Configure settings and click Save
  • Configure API URL (e.g. https://alexa.yourdomain.com) to match 1.8 and 3.1.
  • Configure API Key to match apiKey selected when starting navidrome-alexa in 2.2
  • If everything was configured correctly you should be able to select your device and play music

5. Play

Add music to the Navidrome UI player queue and press play "⏵" button on the widget.

Monitoring

Monitoring

Navidrome-alexa has endpoint metrics exposed via Prometheus/OpenMetrics endpoint at /metrics.
There is also a basic cached healthcheck endpoint at /health If you want to exclude those from public access you can configure a rule to do so:

  alexa.yourdomain.com {
      @exclude not path /health/* /metrics/* /metrics /health 
      reverse_proxy @exclude localhost:4533
  }

Logging

To enable request/response logging for REST endpoints and Alexa client use configuration options logIncomingRequests and logOutgoingRequests. This will leak potentially sensitive data (like authentication tokens) into log files.

Setting up development environment

Building locally

Run in root project folder

go build -o ./build/ ./...

You can also use provided build script to run static checks, tests and build.

Running tests

go test ./...    

Known issues & todo

• Authentication may be tricky and may require authing from a mobile app on the same network first to do CAPTCHA.
• Proper integration with Navidrome vs injected widget
• Better UI for playback controls / progress
• Per-device queue / state
• More control over logging configuration
• Voice commands are likely out of scope (although stop, resume, next, prev are supported if it already has a queue), also there is asknavidrome
• Proper signature validation of incoming /skill requests
• Test Alexa supported formats and if transcoding works/fixes issues, document it
• Multiroom playback, while it does not work with skills out of the box there are potential workarounds to explore