However, it gives an ArrayFire exception when executed:
terminate called after throwing an instance of 'af::exception'
what(): ArrayFire Exception (Internal error:998):
In function cuda::Module common::compileModule(const string&, const std::vector<std::cxx11::basic_string >&, const std::vector<std::__cxx11::basic_string >&, const std::vector<std::cxx11::basic_string >&, bool)
In file src/backend/cuda/compile_module.cpp:276
NVRTC Error(6): NVRTC_ERROR_COMPILATION
Log:
16583636026579851833(258): error: attribute "global" does not apply here
16583636026579851833(259): error: incomplete type is not allowed
16583636026579851833(259): error: expected an identifier
16583636026579851833(259): error: expected an identifier
16583636026579851833(259): error: expected an identifier
16583636026579851833(259): error: expected an identifier
16583636026579851833(259): error: expected an identifier
16583636026579851833(259): error: expected an identifier
16583636026579851833(229): warning #177-D: function "isnan(T) [with T=half]" was declared but never referenced
16583636026579851833(213): warning #177-D: fu
Aborted at 1639583242 (unix time) try "date -d @1639583242" if you are using GNU date
PC: @ 0x7f0271d1e18b gsignal
SIGABRT (@0x9de) received by PID 2526 (TID 0x7f026d4e0000) from PID 2526; stack trace:
@ 0x7f02c5fce631 (unknown)
@ 0x7f02bf1be3c0 (unknown)
@ 0x7f0271d1e18b gsignal
@ 0x7f0271cfd859 abort
@ 0x7f0271fc5911 (unknown)
@ 0x7f0271fd138c (unknown)
@ 0x7f0271fd13f7 std::terminate()
@ 0x7f0271fd137f std::rethrow_exception()
@ 0x557868c17e2b fl::PrefetchDataset::get()
@ 0x557868a3892f _ZZ4mainENKUliE1_clEi
@ 0x5578689ad544 main
@ 0x7f0271cff0b3 __libc_start_main
@ 0x557868a366ce _start
Aborted (core dumped)
The format of the .lst file is as follows (with an empty transcription):
train001 /home/jonah/Downloads/sample.wav 51000
Does anyone know what is going wrong here? Any help would be appreciated.
Question
I am trying to execute the following command to transcribe a .wav file using one of the pretrained models for Flashlight ASR:
/home/jonah/vcpkg/installed/x64-linux/tools/flashlight-cuda/fl_asr_test --am="/media/jonah/Linux Data/Flashlight-speech-recognition/am.bin" --lexicon="/media/jonah/Linux Data/Flashlight-speech-recognition/train_lexicon.txt" –datadir="/media/jonah/Linux Data/Flashlight-speech-recognition/" --test="/media/jonah/Linux Data/Flashlight-speech-recognition/test.lst" --tokens="/media/jonah/Linux Data/Flashlight-speech-recognition/tokens.txt" --emission_dir='' --nouselexicon –show
However, it gives an ArrayFire exception when executed:
The format of the .lst file is as follows (with an empty transcription):
train001 /home/jonah/Downloads/sample.wav 51000
Does anyone know what is going wrong here? Any help would be appreciated.
Additional Context
Running on Ubuntu 20.04 with an RTX 3080.