Add Streaming speech-to-text support for aws-android-sdk-transcribe

[AptFox]

Which AWS Services is the feature request for? aws-android-sdk-transcribe - AWS Transcribe Medical Streaming

Is your feature request related to a problem? Please describe. No, I'd like to use the Android SDK to convert speech to text on the fly. It seems like the SDK only supports a flow involving S3 buckets.

Describe the solution you'd like I'd like to use the Android SDK to convert speech to text on the fly.

Describe alternatives you've considered I'm currently attempting to access the streaming API directly using a websockets library. However, the signature 4 generation is a nightmare.

Additional context N/A

[tao1]

+1

[jeromelesaux]

+1

[MacPower]

+1 hope it will integrated very soon :)

[ryanfurness]

Related? https://github.com/aws-amplify/aws-sdk-android/pull/1759

[AptFox]

I just looked through #1759 and it doesn't seem related to this request. Nothing in it enables streaming transcription.

[Suraj820]

Hi @AptFox am also working on speech to text using aws transcribe i need your help can you help me please..

[AptFox]

Hi @AptFox am also working on speech to text using aws transcribe i need your help can you help me please..

@Suraj820 I cannot.

[hzbcdut]

+1

[VaqibChauhan]

+1

[MKevin3]

We are just getting started down this path and finding any docs on how to use this with Android is near impossible. We also want near real-time transcription meaning web sockets. This will be very short bits of text to transcribe - thinking 8 to 12 words where user is asking for various parts of a complex filter dialog to be filled in based on what they say. I have this working using the android.speech.SpeechRecognizer library.

There will also be an iOS version. We could use Google and iOS specific implementations but then each side would need to maintain the ML aspects of things too. Doing it through Amazon would allow us to have the ML in one place, it can be updated outside the apps, and would offer nearly identical results for both platforms.

The issue is development effort to get web sockets working for both sides. Would rather see this hidden by an Amazon library.

[ryanfurness]

See if this helps you - last year after not having this solved and partnering with some of our AWS team: https://github.com/northwesternmutual/react-native-transcribe

[taouichaimaa]

Any updates on this? and could you please update your docs, the android docs are a nightmare to be honest. I've been browsing your official site for hours but i can't find anything on how to use the android transcribe sdk.

[rogerr99s]

Since java library can do that (implementation "software.amazon.awssdk:transcribestreaming:2.20.0") do you recommend to use that instead of doing synchronous like temporal solution in: https://github.com/aws-amplify/aws-sdk-android/issues/2781 ?

I already implement google cloud speech recognition but we are facing the same issue as @MKevin3 we want to implement the same model in multiple platforms and we want to expand and create new languages

[rogerr99s]

Any news about that? Im getting this error when implementing the java one

type=1400 audit(0.0:58580): avc: denied { read } for name="somaxconn" dev="proc" ino=16755461 scontext=u:r:untrusted_app:s0:c58,c257,c512,c768 tcontext=u:object_r:proc_net:s0 tclass=file permissive=0

[ashirkhan94]

Hi team Any Update on this Same issue for me thanks

[sinyu1012]

Hi, Here is the transcribe streaming code I wrote for Android, you can try it. 🚀 https://github.com/sinyu1012/aws-transcribe-streaming-example-android