[BETA] Emby with Alexa (French)

[DGMayor 88]

I'll play with it more tonight.  I'll also throw some more music in my library (right now I only put the one band, Underworld).

[jhoff80 87]

You are correct in that requesting Nine Inch Nails now works.  I did notice one small step backwards though otherwise (across all artists).

 

After requesting an artist, it lists the first three albums.  There is no more pause between the third album name and "Hear more?".  Then, when I say yes to that question, it errors out.

[Angelblue05 4130]

Thank you, should be fixed now. Let me know if there's anything else

[jhoff80 87]

So now if I say any number to select that album, it hears correctly, but the command it executes is as if I said "yes" and so it lists the next three albums.

[Angelblue05 4130]

Should be fixed now. I was adding a bit more functionality ( adding and removing users from the Emby session ) and I had a naming conflict.

[jhoff80 87]

Yes, fixed.

[jhoff80 87]

Really working great now overall, to the point where I rearranged my bedroom to connect the Echo Dot to good speakers. 

 

Still having trouble with that one album not playing back because of errors where it's trying to transcode for some reason, but everything else seems fairly perfect.  (I mean, Alexa's translation of some of my album titles is bad, but that's not the fault of this skill).

[Angelblue05 4130]

Can you give me an example of a title that Alexa doesn't pronounce right?

 

And I haven't forgotten. It's just really strange. Do you think it's an issue with that particular album? Does it generate a transcode log in emby?

[jhoff80 87]

It's more that Alexa just doesn't understand me saying some of the weirder named albums.  Mer de Noms, for example, is frequently heard as meridian.

 

As for that specific album, I agree that it's really weird.  It seems to play everywhere else.  The transcode log is in the same post as the main log.  The only thing I can see weird about that album versus others is that I tagged it with abnormally high resolution album art (2839x2789, 2MB), not sure if that would be a factor.  Even other albums with 1500x1500 art seem to be playing fine.

 

I did see another weird one with Alexa understanding what I'm saying and it not getting translated properly though.  "Play the album thirteenth step" leads to "I could not find 13rd step".

 

Edit:  Another really weird edge case is for a Clutch song.  When I tell it to "play the song 10001110101" which it correctly recognized as me saying "play the song one zero zero zero one one one zero one zero one", it just returns that there was a problem with the requested skill's response.  I kind of attempted that song only because I suspected it might not work though. 

Edited June 13, 2018 by jhoff80

[Angelblue05 4130]

Hah, @@jhoff80 just read your edit.

 

Try this google form with your edge cases. Additions should be picked up instantly. I plugged it into the skill's title conversion. I figured, there would be quite a bit of edge cases. Now people can try to fix it themselves.

 

https://goo.gl/forms/VgUty6pcSYOPfZH33

 

Just an example, do not add it. This one I will fix within Alexa because I think it confuses it with a choice. This is just to give you the idea.

 

Your incorrect title would be: one zero zero zero one one one zero one zero one

Your title: 10001110101

 

Let me know if you get it to work successfully.

Edited June 13, 2018 by Angelblue05

[jhoff80 87]

Haven't had much time to use this today or yesterday, but yes, that song does get recognized properly now.  When it says what's playing it reads it out as 10 billion (etc.) which is slightly amusing, but yeah it played it fine.

[Angelblue05 4130]

Heh. I guess it would need to say it like 1.0.0...so it doesn't try to read it as a number. Since it's particular to the song, I don't think there's a way to fix it.  But something to think about.

Edited June 15, 2018 by Angelblue05

[Angelblue05 4130]

@@jean-gui

@@popy

 

I've posted the complete translation for Alexa to onesky.

 

Let me know if you are not set up to translate emby projects, I'll add you.

[jhoff80 87]

Found another album that doesn't play back.  Again been tagged with higher resolution album art than usual.  In this case, 2948x3291, which means it's reporting bitrate of the file at 428434 even though it's a 320kbps MP3.  So it looks like what happens is it exceeds the maximum limit for the Alexa player (looks to be set to 384000), and then attempts to transcode, and the transcode seemingly does nothing.

It seems like this is a server issue more than anything else though.

 

Transcode log:

2018-06-16 16:34:10.013
Application version: 3.4.1.16

https://remote.<server>:8097/emby/Audio/60cb1cf45695e3dd058f142b97c55477/hls1/main/0.mp3?PlaySessionId=c2bb4fb77d9048bc9366ceb5572e411e&MaxStreamingBitrate=384000&MaxSampleRate=48000&EnableRedirection=True&TranscodingContainer=mp3&Container=mp3%2Caac&AudioCodec=aac&UserId=a0e98ef148b34a4fb2552434c9119804&TranscodingProtocol=hls&EnableRemoteMedia=True&DeviceId=b1639a25927039311981edbe5e4ef981&api_key=28d0f1b6a8d347ee80935e25d75b9ef9&SegmentContainer=mp3&AudioBitrate=384000

{"Protocol":"File","Id":"60cb1cf45695e3dd058f142b97c55477","Path":"\\\\server\\music\\Nine Inch Nails\\Pretty Hate Machine\\01 - Head Like a Hole.mp3","Type":"Default","Container":"mp3","Name":"01 - Head Like a Hole","IsRemote":false,"ETag":"e490e2dc1ca44fde658aa1031fdaee76","RunTimeTicks":3001989888,"ReadAtNativeFramerate":false,"IgnoreDts":false,"IgnoreIndex":false,"GenPtsInput":false,"SupportsTranscoding":true,"SupportsDirectStream":true,"SupportsDirectPlay":true,"IsInfiniteStream":false,"RequiresOpening":false,"RequiresClosing":false,"RequiresLooping":false,"SupportsProbing":true,"MediaStreams":[{"Codec":"mp3","TimeBase":"1/14112000","CodecTimeBase":"1/44100","DisplayTitle":"MP3 stereo","IsInterlaced":false,"ChannelLayout":"stereo","BitRate":320000,"Channels":2,"SampleRate":44100,"IsDefault":false,"IsForced":false,"Type":"Audio","Index":0,"IsExternal":false,"IsTextSubtitleStream":false,"SupportsExternalStream":false,"Level":0,"IsAnamorphic":false},{"Codec":"mjpeg","Comment":"Cover (front)","TimeBase":"1/90000","CodecTimeBase":"0/1","IsInterlaced":false,"BitDepth":8,"RefFrames":1,"IsDefault":false,"IsForced":false,"Height":2948,"Width":3291,"RealFrameRate":90000,"Type":"EmbeddedImage","AspectRatio":"3291:2948","Index":1,"IsExternal":false,"IsTextSubtitleStream":false,"SupportsExternalStream":false,"PixelFormat":"yuvj420p","Level":-99,"IsAnamorphic":false}],"Formats":[],"Bitrate":428434,"RequiredHttpHeaders":{}}

User policy for Joe: EnablePlaybackRemuxing: True EnableVideoPlaybackTranscoding: True EnableAudioPlaybackTranscoding: True

C:\Users\Joe\AppData\Roaming\Emby-Server\system\ffmpeg.exe -i file:"\\server\music\Nine Inch Nails\Pretty Hate Machine\01 - Head Like a Hole.mp3" -threads 0   -acodec aac -strict experimental -ab 384000 -ac 2 -vn -f segment -max_delay 5000000 -avoid_negative_ts disabled -map_metadata -1 -map_chapters -1 -start_at_zero -segment_time 3  -individual_header_trailer 0 -segment_format mp3 -segment_list_type m3u8 -segment_start_number 0 -segment_list "E:\Emby\transcoding-temp\dfeda7f798d3eaa72d16bdbe99fe209d.m3u8" -y "E:\Emby\transcoding-temp\dfeda7f798d3eaa72d16bdbe99fe209d%d.mp3"

ffmpeg version 4.0 Copyright (c) 2000-2018 the FFmpeg developers
  built with gcc 7.3.0 (GCC)
  configuration: --enable-gpl --enable-version3 --enable-sdl2 --enable-bzlib --enable-fontconfig --enable-gnutls --enable-iconv --enable-libass --enable-libbluray --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libtheora --enable-libtwolame --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libzimg --enable-lzma --enable-zlib --enable-gmp --enable-libvidstab --enable-libvorbis --enable-libvo-amrwbenc --enable-libmysofa --enable-libspeex --enable-libxvid --enable-libaom --enable-libmfx --enable-amf --enable-ffnvcodec --enable-cuvid --enable-d3d11va --enable-nvenc --enable-nvdec --enable-dxva2 --enable-avisynth
  libavutil      56. 14.100 / 56. 14.100
  libavcodec     58. 18.100 / 58. 18.100
  libavformat    58. 12.100 / 58. 12.100
  libavdevice    58.  3.100 / 58.  3.100
  libavfilter     7. 16.100 /  7. 16.100
  libswscale      5.  1.100 /  5.  1.100
  libswresample   3.  1.100 /  3.  1.100
  libpostproc    55.  1.100 / 55.  1.100
Input #0, mp3, from 'file:\\server\music\Nine Inch Nails\Pretty Hate Machine\01 - Head Like a Hole.mp3':
  Metadata:
    title           : Head Like a Hole
    artist          : Nine Inch Nails
    track           : 1/11
    album           : Pretty Hate Machine
    disc            : 1/1
    TMED            : CD

[jhoff80 87]

More to the point for this app, I've noticed when I ask to play an artist, it lists 3 albums.  If I ask to play a song and it's suggesting different options, it seems to list every option the search has found.  To me, 3 seems a little too limited for the albums.  But when it's listing off like 9 different songs before I can respond, that feels like too much.  I think it'd make sense for these to meet somewhere in the middle.  Maybe list 5 options in both scenarios?  Just an idea though.

[Angelblue05 4130]

You know you can interrupt Alexa by saying Alexa. If you hear what you want, just go Alexa, # and then she'll proceed. You don't have to let her go through the entire list before saying something.

 

But I hear you. I'll make adjustments.

Edited June 16, 2018 by Angelblue05

[Angelblue05 4130]

I'll also look into the playing error. I'm not currently creating a device profile, hopefully that will resolve any corner cases.

[jean-gui 2]

@@Angelblue05 Thanks, I seem to be able to edit the translation. One question though, what is the syntax of those strings? I would guess that the first piece (before the first /) is the key and shouldn't be translated, can you confirm that?

[Angelblue05 4130]

@@jean-gui

 

I take it to mean you cannot see my description for translation? No problem. I've added notes to the individual sentences where need be. Everything needs to be translated (except if the individual note states otherwise). I fixed the duplicates just now.

 

Since this is more complex than one value one string i.e. I want to play a song. Play a song. Put on a song.

The translation entry would look like that I want to play song./Play a song./Put on a song.

You could have less entries or more entries, as long as they are separated by one / if there's more than one.

 

Words in brackets need to remain the same, but they can be moved around to accommodate your translation. Words in bracket for intents.json have additional information. i.e. {Play:play_terms} a song. This just indicates the "word bucket" used. You'll find the entry play_terms:Play with all the possible words that refer to that word bucket.

 

That's it. Let me know if it's not clear. If something is too difficult to translate because of the format then I'll try to think of another way to break down the Alexa model. 

Edited June 19, 2018 by Angelblue05

[jean-gui 2]

@@Angelblue05

OK thanks for the explanation. I also found it on the translation site .

I have started translating the strings(I've added a few comments as well) and will continue tomorrow.

[Angelblue05 4130]

Sorry, I should have told you I've updated the OP with the translation info.

 

I've edited the description. Duplicates are normal. It's just because they belong to different word buckets, but the words are the same.

 

Merci beaucoup.

Edited June 19, 2018 by Angelblue05

[jean-gui 2]

I'm done translating https://oshotim.oneskyapp.com/collaboration/translate/project/project/147181/language/5#/ but I'm not sure about the last blocks containing several words in brackets as it can be hard to grasp their intended use. Some example or explanation would help a lot.

 

I also see that someone else took care of https://oshotim.oneskyapp.com/collaboration/translate/project/project/147183/language/5#/, great

Edited June 21, 2018 by jean-gui

[Angelblue05 4130]

Replied to your comments.

 

All done? How was it?

Edited June 23, 2018 by Angelblue05

[jean-gui 2]

As I mentioned briefly in some comments on the translation platform, I found the translation UI not very usable. First, browser search works poorly (at least on my computer). You can't search for translation of a specific key, so you have to manually check all the translations until you find what you're looking for. Also, using new lines instead of / as an alternative delimiter would make it easier to read.

 

Call me old school, but I feel like being able to edit files directly  would have made the process easier. As a comparison, I've also translated Kanzi (Alexa for Kodi, https://github.com/m0ngr31/kanzi/pull/267) and it felt easier.

 

Anyway, all done, I think! There will likely be issues, notably with the strings containing a lot of bracketed words, but it will be easier to figure them out when testing. Let me know if there are some more things to do.

Edited June 23, 2018 by jean-gui

[Angelblue05 4130]

Well it is the first time I ever translate such a complex project. So this is really trial error as you are the first person to go through it.

 

I'lI try to see if it's possible to have new lines. I don't have a problem searching string. You have to use the string key, not the random string number (no idea why the string key only appears on the right side pane, under suggestion).