search

flashlight / wav2letter

Facebook AI Research's Automatic Speech Recognition Toolkit
https://github.com/facebookresearch/wav2letter/wiki
Other
6.39k stars 1.01k forks source link

what(): loadSound: unknown format or could not open stream #360

Closed cogmeta closed 4 years ago

cogmeta commented 5 years ago

I followed the librispeech recipe and I get this

terminate called after throwing an instance of 'std::runtime_error' what(): loadSound: unknown format or could not open stream Aborted at 1563884075 (unix time) try "date -d @1563884075" if you are using GNU date PC: @ 0x7faea54da428 gsignal SIGABRT (@0x3e8000169a7) received by PID 92583 (TID 0x7faef8770640) from PID 92583; stack trace: @ 0x7faef7cfd390 (unknown) @ 0x7faea54da428 gsignal @ 0x7faea54dc02a abort @ 0x7faec26a10a5 gnu_cxx::verbose_terminate_handler() @ 0x7faec2612436 cxxabiv1::terminate() @ 0x7faec2612481 std::terminate() @ 0x7faec260fee3 cxa_throw @ 0x5d2951 w2l::loadSound<>() @ 0x5d2b10 w2l::loadSound<>() @ 0x5c626a w2l::W2lListFilesDataset::getLoaderData() @ 0x5bf564 w2l::W2lDataset::getFeatureData() @ 0x5c0b19 w2l::W2lDataset::getFeatureDataAndPrefetch() @ 0x5c0e4e w2l::W2lDataset::get() @ 0x4804a4 _ZZ4mainENKUlSt10shared_ptrIN2fl6ModuleEES_IN3w2l17SequenceCriterionEES_INS3_10W2lDatasetEES_INS0_19FirstOrderOptimizerEES9_ddbiE3_clES2_S5_S7_S9_S9_ddbi.constprop.11256 @ 0x41b525 main @ 0x7faea54c5830 libc_start_main @ 0x47bcb9 _start @ 0x0 (unknown) Aborted (core dumped)

cogmeta commented 5 years ago

Training config for Librispeech using Gated ConvNets Replace [...] with appropriate paths

--runname=librispeech_conv_glu --rundir=models --tokensdir=./data/librispeech --archdir=./config/seq2seq_tds --listdata=true --train=./data/librispeech/train-clean-100.lst,./data/librispeech/train-clean-360.lst,./data/librispeech/train-other-500.lst --valid=dev-clean:./data/librispeech/dev-clean.lst,dev-other:./data/librispeech/dev-other.lst --lexicon=./data/librispeech/librispeech-train+dev-unigram-10000-nbest10.dict --arch=network.arch --tokens=librispeech-train-all-unigram-10000.vocab-filtered --lexicon=./data/librispeech/librispeech-train+dev-tokens.dict --arch=network.arch --tokens=tokens.txt --criterion=asg --lr=0.6 --lrcrit=0.006 --linseg=1 --momentum=0.8 --maxgradnorm=0.2 --replabel=2 --surround=| --onorm=target --sqnorm=true --mfsc=true --nthread=6 --batchsize=4 --transdiag=4

cogmeta commented 5 years ago

head -5 ./data/librispeech/train-clean-100.lst train-clean-100-103-1240-0000 /home/ubuntu/data/wav2letter/recipes/librispeech/data/src/LibriSpeech/train-clean-100/103/1240/103-1240-0000.flac 14085.0 chapter one missus rachel lynde is surprised missus rachel lynde lived just where the avonlea main road dipped down into a little hollow fringed with alders and ladies eardrops and traversed by a brook train-clean-100-103-1240-0001 /home/ubuntu/data/wav2letter/recipes/librispeech/data/src/LibriSpeech/train-clean-100/103/1240/103-1240-0001.flac 15945.0 that had its source away back in the woods of the old cuthbert place it was reputed to be an intricate headlong brook in its earlier course through those woods with dark secrets of pool and cascade but by the time it reached lynde's hollow it was a quiet well conducted little stream train-clean-100-103-1240-0002 /home/ubuntu/data/wav2letter/recipes/librispeech/data/src/LibriSpeech/train-clean-100/103/1240/103-1240-0002.flac 13945.0 for not even a brook could run past missus rachel lynde's door without due regard for decency and decorum it probably was conscious that missus rachel was sitting at her window keeping a sharp eye on everything that passed from brooks and children up

cogmeta commented 5 years ago

ls -al "/home/ubuntu/data/wav2letter/recipes/librispeech/data/src/LibriSpeech/train-clean-100/103/1240/103-1240-0000.flac" -r--r--r-- 1 ubuntu ubuntu 255398 Jul 8 2014 /home/ubuntu/data/wav2letter/recipes/librispeech/data/src/LibriSpeech/train-clean-100/103/1240/103-1240-0000.flac

tlikhomanenko commented 5 years ago

Hi @cogmeta,

Seems you have a trouble with libsnd. Could you provide details how you installed it? Could you try to run this in the docker container?

nhha1602 commented 5 years ago

I got the same issue. I installed libsnd by following instruction:

git clone git://github.com/erikd/libsndfile.git ./autogen.sh ./configure --enable-werror make make check sudo make install

I dont use docker.

cogmeta commented 5 years ago

@nhha1602 Thanks. Will try it and report back.

nhha1602 commented 5 years ago

I tried to re-build libsndfile with opus and it is OK.

samehraban commented 5 years ago

I tried to re-build libsndfile with opus and it is OK.

How did you do that? I'm facing same problem and it seems libsndfiles is configured to read opus/flac by default.

kksiddharth commented 4 years ago

Did anyone find a solution to this problem.... I'm facing the same error when i try to train the acoustic model of the LibriSpeech clean tutorial, I tried out @nhha1602 suggestion to rebuild libsndfile with opus, https://github.com/xiph/opus, but to no avail as the error is still persisting, any help on this issue would be of great help, thanks

kksiddharth commented 4 years ago

Hey guys, I'm really sorry, but yea @nhha1602 was right, I just had to rebuild wav2letter after rebuilding libsndfile with opus, and the issue went away, one way, correct me if I'm wrong, i guess this can be detected is to make test after building wav2letter, all the cases were passing except SoundTest for me when i built libsndfile without opus, and after rebuilding with opus it passed. Thanks