Speech recognition from audio file?

[marc.ric...@gmail.com]

Hi everyone,

I'd like to generate the following automation:

When I hit a homescreen shortcut, automation should record an audio file and send it to a predefined recipient by mail.

This is already working, using this setup:

When using this frequently, it might get quite annoying and confusing to have several mails with the same subject and message in your inbox, as set in the "Gmail send" step. It would be much better, if the recorded audio could be transcribed to text by speech recognition and be used for subject and/or message.

I could not find a way to achieve this; does anybody have an idea on this or another convenient method?

If this is not possible currently: Feature Request! :)

[abin mn]

You could use the speech recognition block to do this 

[marc.ric...@gmail.com]

Hi Abin,

how exactly? I understand that the speech recognition block just recognizes speech and saves it as an array of written text, which can be used by other blocks then.

But this way, I do not have an audio file anymore to attach to that mail.

[Henrik "The Developer" Lindqvist]

Why would you need both, either send the audio file, or the transcript?

[marc.ric...@gmail.com]

As I explained in the first message: When using this automation often, it might get quite annoying and confusing to have several mails with the same subject and message in your inbox, as set in the "Gmail send" step. It would be much better, if the recorded audio could be transcribed to text by speech recognition and be used for subject and/or message so you have a preview of what's in the audio file, just by having a look at the mail's text.

But since speech recognition isn't much accurate, and maybe there's more to hear in the audio file but only the text, the attached audio file is needed also. Think of a musician, who wants to sing an idea for a new song and have that sent to his mailbox: He needs the melodys, timings and so on, too. Not only the text. Stars like Madonna are known for doing so by calling their own answering machine for this. 

[abin mn]

You could use speech recognition block to add the subject and then record your audio

--
You received this message because you are subscribed to a topic in the Google Groups "Automate" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/automate-user/LSAmYKvn25E/unsubscribe.
To unsubscribe from this group and all its topics, send an email to automate-use...@googlegroups.com.
To post to this group, send email to automa...@googlegroups.com.
Visit this group at https://groups.google.com/group/automate-user.
To view this discussion on the web visit https://groups.google.com/d/msgid/automate-user/cda319b5-e2ad-41c2-b853-30a34b0242ed%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[marc.ric...@gmail.com]

That would make it necessary to make two recordings; one for the SR of the subject and a second for the actual audio file. That is a good idea and I already had it, too. It serves as a good workarround at least. But it still would be even cooler if Speech Recognition of an audio file would be possible. It would broaden the use cases of Automation for several cool functions I have in my mind - queue.

[Henrik "The Developer" Lindqvist]

Sadly the Speech recognition in Android doesn't support using an file as audio input:
http://developer.android.com/reference/android/speech/SpeechRecognizer.html

You may be able to use some online speech recognition API:
https://cloud.google.com/speech/

[marc.ric...@gmail.com]

Thanks for that investigation anyways!

Even a negative result serves, since I now know that there's no sense in hanging on that plan.

Thank you for your efford!