Google Speech API via gRPC recognizes first few words only. Tested using several short (<10s) audio files in several (wav, flac) formats.
I am trying to get Google's speech API to work with Delphi using gRPC. (Calling the Speech API from Delphi using JSON works OK, but does not support streaming)
As Google does not support Delphi as gRPC client, I use Grijjy's protocolbuffers for Serializing/Deserializing audio request/response, and Grijjy's scalable sockets for the http/2 request.
As I think the code is too large to share here, I created a MVCE. A zip file containing source code (Delphi 10.3), executable and audio file can be found here. The audio file contains the text "How old is the Brooklyn Bridge", which is flawlessly recognized when calling the Speech API via JSON.
To run the executable, only a Google API key is necessary. Necessary support file nghttp2.dll is included. After sending a request, a messagebox is displayed containing the deserialized response. I checked this is the complete Google response by checking response data size. The response only shows "how old", where I would have expected a full recognized sentence. No errors are returned, confidence factor is also returned.
To compile the source code, Grijjy's code as mentioned above must downloaded from Github and the folder(s) containing the downloaded code made available to the Delphi project.
Also, for Google to accept the request, in grijjy.http, line 1215 needs to modified from
FInternalHeaders.AddOrSet('host', FURI.Host);
to
FInternalHeaders.AddOrSet(':authority', FURI.Host);
(I failed to create a pull request for this earlier today, will try again later)
During my debugging efforts, I also checked response headers (200 OK), also tried other files (same issue).
For now, I do not know how to continue. Any ideas regarding how to get find the cause (or even better, a solution:-) of this issue would be highly appreciated.
Related
Currently, I am building a javascript app that provides OneDrive filesystem support for chromeOS. Unfortunately, I encounter an issue when downloading a file via the OneDrive API due to preflights failing.
Deeply buried in the Microsoft API docs there is a mention of a workaround for CORS request for javascript apps, because preflights will fail. In order to circumvent a 302 redirect, a file download has to be executed in two steps: first get the download url, then download the file via that url. This works fine as a simple GET request doesn't require a preflight check.
Back to the filesystem. According to the chrome.fileSystemProvider API docs:
The results must be returned in chunks by calling successCallback several times.
This can be achieved by using the Range header on the GET request to the download-url, which actually is supported by OneDrive API using partial range downloads. However, now the request isn't simple anymore and therefore requires a preflight check, which ultimately fails.
So basically, the workaround failed because the problem the workaround tried to work around is still present in the workaround.
How can I overcome this problem? I obviously don't want to funnel the request through a proxy because of bandwidth and privacy issues. Can I perhaps simulate the Range header by downloading the entire file first and then serving it in byte-size chunks to the fileSystemProvider?
I came across a question asked almost 5 years ago with a very similar issue, but unfortunately it didn't get me any further.
In an earlier post from a few minutes ago, I asked a "general" question regarding creating general webservices in vaadin: How can one create webservices in Vaadin 12?
However, one specific unique case that I mainly need to support is the uploading via https of large (eg ~400MB) bytestream objects that would presumably be sent to Vaadin via an https "post" command (with the paylod being provided I presume in raw binary format as a bytestream.) I saw that Vaadin has built-in support for uploading files (which is essentially a post command of a bytestream, I presume?) and then I saw a reference to StreamReceiver here: https://vaadin.com/docs/v12/flow/advanced/tutorial-stream-resources.html
which seems to sound like a custom file importer, but I couldn't find any (simple & more-or-less complete) examples on how to use it. Ideally, a quick few lines of Java to show the "receiving" of the bytestream and a few quick lines (ideally in Java) which "posts" to the receivestream's url would be all that's needed to show how this manual upload of bytes can be accomplished in Vaadin. (In DropWizard & Jersey, I can find such examples reasonably easily, but I'm not sure how to gain that level of control in Vaadin.)
(Very very minor bonus: is there a size limit to the post command? eg, can a bytestream of over say ~4GB be sent and received?)
In Vaadin the Upload API is optimised for streaming into File (unlike handling the stream as in Servlet and JAX-RS API). One way is to first stream to a temp file and then when the file is fully on the server side, handle the data from temp file.
Alternatively you can use Flow Viritin add-on and a helper class UploadFileHandler, which give you and API where you read the contents from InputStream, in same way as with Servlet API. See a usage example is in this test.
This isn't a first time this is asked and I actually have a more verbose blog draft about this subject. I'll add a link to that once I get that published.
What's the correct way of transferring media (photos or movies) using Worklight Adapters?
I sent a photo via the adapter and got the error: form too large, exceed the maximum size...
I read I need to change the form size through the Jetty
but the server I'll deploy the app won't be a jetty so what shell i do?
Thanks!
Please see topic Uploading large (and binary) files to Worklight adapter.
Basically, Worklight does not have the equivalent to an HTTP POST mechanism that allows you to transfer arbitrarily large chunked data. For large files of unknown sizes (photos, video, audio) you'll need to upload the file to the server outside the Worklight adapter framework. For example you could simply post it to a web server you have configured. In my case (in the above referenced answer) I needed to create an entire client-server mechanism to negotiate a port and key, start listening on that port, then accept requests and ensure the posting client passes the key as authorization to transfer the secure data.
Hopefully IBM will provide a formal service for this in a future release.
Adapters do not work with html forms, they work with data.
You will need to convert your image to base64 and submit as a adapter invocation parameter.
Having more information regarding what exactly you're trying to achieve might be helpful.
How do I play a track from a SoundCloud URL, which, for example, I got from the xml response from a query
<stream-url>https://api.soundcloud.com/tracks/31164607/stream</stream-url>
I should have thought that it would have been as easy as:
https://api.soundcloud.com/tracks/31164607/stream&client_id=my_client_id
yet I get
<error>401 - Unauthorized</error>
All I want to do is consume it in a Silverlight MediaElement, so all I need is set some url to the MediaElement's Source property.
I've checked an application that I wrote about 2 years ago, and THEN, accessing the stream url was as easy as this for a public track:
http://api.soundcloud.com/tracks/18163056/stream&consumer_key=MY_CONSUMER_KEY
however this no longer seems to work.
For example, all I had to do then in C# was:
MediaElement me = new MediaElement();
me.Source= new Url("http://api.soundcloud.com/tracks/18163056/stream&consumer_key=MY_CONSUMER_KEY");
me.Play();
Any hints would be appreciated.
I had a reply on a Microsoft forum that seems to imply that SoundCloud might not be possible to stream to Windows 8 Metro devices without consuming the whole stream before playback starts - which is quite worrying and would seem to imply that to make authentication possible, it would have to be done entirely in the url querystring insterad of using the header:
(The following reply is the answer to the following question: 'I am able to access an audio stream by http using the MediaElement, however I need to access it via https in which I need to add the oAuth info to the header of the initial request.
How is this done when using a MediaElement, and if it cannot be done, what is the workaround for consuming an audio feed in Metro 8 that requires header authentication to stream?')
"Direct access to the underlying network stream is not currently permitted by the MediaElement. Because of this there is currently no way to modify the header of the HTTP request to include any additional authentication information. That said, you do have control over the URL. You could theoretically setup an HTTP proxy service that translated the HTTP GET request parameters into the necessary oAuth credentials. Keep in mind that this is just a theoretical workaround. You may find different behavior in practice. Another theoretical workaround would be to handle the oAuth yourself via a raw stream socket and pass the retuned media data to the MediaElement via "Set Source" and a "Random Access Stream". Please keep in mind that this method has major limitations. in order to use a "Random Access Stream" with the ME you need to make sure all of the data is available before passing it to the ME."
The proxy service is not scalable for an application that is merely distributed for free as every stream would need to come via the proxy. And the raw stream socket, although getting around this, would mean that playback could not start until the whole file had downloaded - and this goes against all current UX (User Experience) guidelines.
So once again, if anyone has any tips, or info about how the whole authentication thing can be achieved in a querystring instead of using headers, I'd appreciate it!
I'm a little confused about whether you're referring to a public or a private track? If it's a public track, then you shouldn't need to send any authentication information, just your client id.
When I request https://api.soundcloud.com/tracks/31164607/stream?client_id=YOUR_CLIENT_ID then I get a 302 redirect to the proper mp3 stream.
Remember, adding parameters to a URL must start with a ? not &. This could (more than likely) be the reason why you are getting a 401 (SC is not picking up the client_id).
After authentication the link like this
http://api.soundcloud.com/tracks/103229681/stream?consumer_key=d61f17a08f86bfb1dea28539908bc9bf
is working fine. I am using Action Script.
I'm following up on Tom's reply because he calls attention to url character specificity. My HTTP requests randomly started failing today, and I was prefacing my client_Id with a ?. As soon as I changed that single ? to &, it started working. So in my case, SC wasn't picking up my client_Id because I used the wrong character. I think depending on where in the request we're talking about specifically, it's worth noting that differences between ? and & do make a difference.
I've a file I want to send to the ebay system to support the LMS.
All the samples I've found include the use of the API, but the environment I'm working in doesn't have the ability to use it (the api).
So I'm forced to send the file with an HTTP post. But the doc's seem lacking.
Has anyone constructed/found an example of a HTTP post that will send a given file.
EDIT:
Oh.. what I see in the samples I have found is an area that seems it's supposed to have the data, but in the sample, there's nothing I'd consider real data.
Are you talking about the file transfer service or the bulk upload service? Don't you just generate an xml document and post the url like in this example:
http://developer.ebay.com/DevZone/file-transfer/CallRef/uploadFile.html#Samples