That’s really not a lot, but properly re-using checkpoint should not yield empty audio I think. Can you make sure you pass a language model, and use releases values for lm_alpha
and lm_beta
.
i was using tensorflow 1.14 so upon running got " excenption handler got a new exception".
downgraded tensorflow to 1.13.1 and now new error is coming.
error: The following untracked working tree files would be overwritten by checkout: .cardboardlint.yml .compute .gitattributes .github/lock.yml .gitignore .pylintrc .taskcluster.yml .travis.yml CODE_OF_CONDUCT.md DeepSpeech.py Dockerfile GRAPH_VERSION ISSUE_TEMPLATE.md LICENSE VERSION bazel.patch bin/benchmark_nc.py bin/benchmark_plotter.py bin/gpu_usage_chart bin/gpu_usage_plot bin/graphdef_binary_to_text.py bin/import_aishell.py bin/import_cv.py bin/import_cv2.py bin/import_fisher.py bin/import_gram_vaani.py bin/import_ldc93s1.py bin/import_librivox.py bin/import_lingua_libre.py bin/import_swb.py bin/import_ted.py bin/import_timit.py bin/import_ts.py bin/import_voxforge.py bin/ops_in_graph.py bin/run-ldc93s1.sh bin/run-tc-ldc93s1_checkpoint.sh bin/run-tc-ldc93s1_new.sh bin/run-tc-ldc93s1_singleshotinference.sh bin/run-tc-ldc93s1_tflite.sh data/alphabet.txt data/smoke_test/LDC93S1.txt data/smoke_test/LDC93S1.wav data/smoke_test/LDC93S1_pcms16le_1_8000.wav data/smoke_test/LDC93S1_pcms16le_2_44100.wav data/smoke_test/vocab.pruned.lm data/smoke_test/vocab.trie data/ted/.gitkeep doc/DeepSpeech.rst doc/Geometry.rst doc/Makefile doc/ParallelOptimization.rst doc/audioTranscript.png doc/conf.py doc/index.rst doc/make.bat evaluate.py evaluate_tflite.py examples/ffmpeg_vad_streaming/README.MD examples/ffmpeg_vad_streaming/index.js examples/ffmpeg_vad_streaming/package.json examples/mic_vad_streaming/mic_vad_streaming.py examples/mic_vad_streaming/requirements.txt examples/net_framework/.gitignore examples/net_framework/DeepSpeechWPF/App.config examples/net_framework/DeepSpeechWPF/App.xaml examples/net_framework/DeepSpeechWPF/App.xaml.cs examples/net_framework/DeepSpeechWPF/DeepSpeech.WPF.csproj examples/net_framework/DeepSpeechWPF/DeepSpeech.WPF.sln examples/net_framework/DeepSpeechWPF/MainWindow.xaml examples/net_framework/DeepSpeechWPF/MainWindow.xaml.cs examples/net_framework/DeepSpeechWPF/Properties/AssemblyInfo.cs examples/net_framework/DeepSpeechWPF/Properties/Resources.Designer.cs examples/net_framework/DeepSpeechWPF/Properties/Resources.resx examples/net_framework/DeepSpeechWPF/Properties/Settings.Designer.cs examples/net_framework/DeepSpeechWPF/Properties/Settings.settings examples/net_framework/DeepSpeechWPF/packages.config examples/nodejs_wav/Readme.md examples/nodejs_wav/index.js examples/nodejs_wav/package.json examples/vad_transcriber/audioTranscript_cmd.py examples/vad_transcriber/audioTranscript_gui.py examples/vad_transcriber/requirements.txt examples/vad_transcriber/wavSplit.py examples/vad_transcriber/wavTranscriber.py examples/vad_transcriber/wavTranscription.md images/LSTM3-chain-backwards.png images/LSTM3-chain.png images/Parallelism.png images/rnn_fig-624x548.png images/usage.gif native_client/Android.mk native_client/BUILD native_client/Makefile native_client/alphabet.h native_client/args.h native_client/client.cc native_client/ctcdecode/COPYING native_client/ctcdecode/LICENSE.paddlepaddle native_client/ctcdecode/LICENSE.parlance native_client/ctcdecode/Makefile native_client/ctcdecode/__init__.py native_client/ctcdecode/build_common.py native_client/ctcdecode/ctc_beam_search_decoder.cpp native_client/ctcdecode/ctc_beam_search_decoder.h native_client/ctcdecode/decoder_utils.cpp native_client/ctcdecode/decoder_utils.h native_client/ctcdecode/numpy.i native_client/ctcdecode/output.h native_client/ctcdecode/path_trie.cpp native_client/ctcdecode/path_trie.h native_client/ctcdecode/scorer.cpp
Please move or remove them before you switch branches.
Aborting
Can you explain ?
I’m sorry, I don’t understand what you are doing here …