Speech to Text Toy Data Could Not Be Downloaded

[Keram-Yasin]

I was evaluating the speech-to-text feature of OpenNMT with this getting started example: https://opennmt.net/OpenNMT-py/legacy/speech2text.html However, the toy training data could not be downloaded with this command given in the example: wget -O data/speech.tgz http://lstm.seas.harvard.edu/latex/speech.tgz;

The response of the above command is: $ wget -O ./speech.tgz http://lstm.seas.harvard.edu/latex/speech.tgz; --2024-01-20 11:45:53-- http://lstm.seas.harvard.edu/latex/speech.tgz Resolving lstm.seas.harvard.edu (lstm.seas.harvard.edu)... 54.172.197.41 Connecting to lstm.seas.harvard.edu (lstm.seas.harvard.edu)|54.172.197.41|:80... connected. HTTP request sent, awaiting response... 301 Moved Permanently Location: https://lstmvis.vizhub.ai/ [following] --2024-01-20 11:45:57-- https://lstmvis.vizhub.ai/ Resolving lstmvis.vizhub.ai (lstmvis.vizhub.ai)... 169.63.55.253 Connecting to lstmvis.vizhub.ai (lstmvis.vizhub.ai)|169.63.55.253|:443... failed: Connection timed out. Retrying.

--2024-01-20 11:48:10-- (try: 2) https://lstmvis.vizhub.ai/ Connecting to lstmvis.vizhub.ai (lstmvis.vizhub.ai)|169.63.55.253|:443... failed: Connection timed out. Retrying.

--2024-01-20 11:50:22-- (try: 3) https://lstmvis.vizhub.ai/ Connecting to lstmvis.vizhub.ai (lstmvis.vizhub.ai)|169.63.55.253|:443...

[vince62s]

Just bear in mind we do not support speech since v2.Not sure what you want to do but maybe this is not the best toolkit to use for you.

[Keram-Yasin]

The Issue still exists. The command downloaded the html page of https://lstmvis.vizhub.ai/ as .tgz file. The file can not be extracted.

@vince62s I had no idea about that. I thought the OpenNMT-py is also good at translating speech-to-text. Any replacement for the absence since v2 from OpenNMT?

[vince62s]

for speech to text, search for "faster whisper"