adroitanandai / indian-accent-speech-recognition Goto Github PK

Hi @AdroitAnandAI ! this is some impressive work, and it would be great to make it more usable by releasing model checkpoints as well... the current protobuf model files are harder to use with newer DeepSpeech releases, but checkpoints can be more easily re-formatted

Very good article and I hope model should also be good. However, not able to run it due to some error. I have downloaded your model which you had put in this repo. Please suggest:

deepspeech --model output_graph_indian_test.pbmm --lm models/lm.binary --audio chunk02_speaker0_16.wav --alphabet models/alphabet.txt --trie models/trie

TensorFlow: v1.12.0-10-ge232881
DeepSpeech: v0.4.1-0-g0e40db6
Data loss: Corrupted memmapped model file: output_graph_indian_test.pbmm Invalid directory offset
Traceback (most recent call last):
  File "/anaconda3/envs/berttorchenv_dev/bin/deepspeech", line 8, in <module>
    sys.exit(main())
  File "/anaconda3/envs/berttorchenv_dev/lib/python3.6/site-packages/deepspeech/client.py", line 80, in main
    ds = Model(args.model, N_FEATURES, N_CONTEXT, args.alphabet, BEAM_WIDTH)
  File "/anaconda3/envs/berttorchenv_dev/lib/python3.6/site-packages/deepspeech/__init__.py", line 14, in __init__
    raise RuntimeError("CreateModel failed with error code {}".format(status))
RuntimeError: CreateModel failed with error code 15

Hi,
Thank you for the article and the model. But for me, when I run the command given in the doc, I am getting
deepspeech: error: ambiguous option: --lm could match --lm_alpha, --lm_beta
and on adding lm-alpha like so
deepspeech --model speech/output_graph.pbmm --lm speech/lm.binary --trie speech/trie --audio content/8455-210777-0068.wav,
I am getting
deepspeech: error: argument --lm_alpha: invalid float value: 'speech/lm.binary'.
My version for Deepspeech is 0.8.2.
I am kinda new at this and any help will be appreciated.
Thank you

Hi,
I'm trying to access your pre-trained model. But it says the "file is in owner's trash". Can you please re-upload it again?
Thanks

hello sir "output_graph.pbmm" is currently not available in your reposetory can you please make it availabe or please provide the file as it is pretrained model and must need for this please accept this request

I am assuming that the code and model provided in the public repository is available for reuse in other applications and personal projects but It would be great to have a licence such as MIT added to the repository to help in making the decision easier.

Greetings ,

The repo is awesome and also thanks for google colab implementation, it's easy to use , but after testing the repo with ease , i got the output from the .wav file as a text (as mentioned), but i am also curious whether can i do a voice clone of my own voice in real time on the generated text in an indian accent?

Any help/guidance would be really helpfull,
Thanks,
satyam.

Hello,
Loved your article and your work. Can you provide trained model along with access to raw data for references.
Really appreciate it.