#!/bin/bash

. ./cmd.sh ## You'll want to change cmd.sh to something that will work on your system.
           ## This relates to the queue.
. ./path.sh # so python3 is on the path if not on the system (we made a link to utils/).a

# This is a shell script, but it's recommended that you run the commands one by
# one by copying and pasting into the shell.


# Download the corpus and prepare parallel lists of sound files and text files
# Divide the corpus into train, dev and test sets
local/sprak_data_prep.sh  || exit 1;
utils/fix_data_dir.sh data/train || exit 1;

# Perform text normalisation, prepare dict folder and LM data transcriptions
local/copy_dict || exit 1;

utils/prepare_lang.sh data/local/dict "<UNK>" data/local/lang_tmp data/lang || exit 1;

# Now make MFCC features.
# mfccdir should be some place with a largish disk where you
# want to store MFCC features.
mfccdir=mfcctest


# Extract mfccs
# p was added to the rspecifier (scp,p:$logdir/wav.JOB.scp) in make_mfcc.sh because some
# wave files are corrupt
# Will return a warning message because of the corrupt audio files, but compute them anyway
# If this step fails and prints a partial diff, rerun from sprak_data_prep.sh

steps/make_mfcc.sh --nj 10 --cmd $train_cmd data/test exp/make_mfcc/test test mfcc || exit 1;
steps/make_mfcc.sh --nj 10 --cmd $train_cmd data/train exp/make_mfcc/train mfcc || exit 1;


# Compute cepstral mean and variance normalisation
steps/compute_cmvn_stats.sh data/test exp/make_mfcc/test mfcc || exit 1;
steps/compute_cmvn_stats.sh data/train exp/make_mfcc/train mfcc || exit 1;



# Repair data set (remove corrupt data points with corrupt audio)

utils/fix_data_dir.sh data/test || exit 1;
utils/fix_data_dir.sh data/train || exit 1;


# Train LM with irstlm
#creates 3g or 4g dictionary and importantly G.fst
#local/train_irstlm.sh data/local/transcript_lm/transcripts.uniq 3 "3g" data/lang data/local/train3_lm &> data/local/3g.log &
local/train_irstlm.sh data/local/transcript_lm/transcripts.uniq 4 "4g" data/lang data/local/train4_lm &> data/local/4g.log || exit 1;

#speed test only 120 utterances per speaker
utils/subset_data_dir.sh --per-spk data/test 120 data/test120_p_spk || exit 1;


# Train monophone model on short utterances  AFTER THIS ONE CAN SEE THE ALIGNMNT BETWEEN FRAMES AND PHONES USING COMMAND SHOW_ALIGNMENTS
steps/train_mono.sh --nj 10 --cmd "$train_cmd" data/train data/lang exp/mono || exit 1;

# Ensure that LMs are created

utils/mkgraph.sh data/lang_test_4g exp/mono exp/mono/graph_4g || exit 1;

# Ensure that all graphs are constructed


steps/decode.sh --config conf/decode.config --nj 10 --cmd "$decode_cmd" \
  exp/mono/graph_4g data/test120_p_spk exp/mono/decode || exit 1;

# Get alignments from monophone system.
steps/align_si.sh --nj 10 --cmd "$train_cmd" \
  data/train data/lang exp/mono exp/mono_ali || exit 1;

# train tri1 [first triphone pass]
# steps/train_deltas.sh --boost-silence 1.25 --cmd "$train_cmd" \
steps/train_deltas.sh --cmd "$train_cmd" \
  5800 96000 data/train data/lang exp/mono_ali exp/tri1|| exit 1;



#make graph
utils/mkgraph.sh data/lang_test_4g exp/tri1 exp/tri1/graph_4g || exit 1;

steps/decode.sh --config conf/decode.config --nj 10 --cmd "$decode_cmd" \
  exp/tri1/graph_4g data/test120_p_spk exp/tri1/decode_test120_p_spk || exit 1;




steps/align_si.sh --nj 10 --cmd "$train_cmd" \
  data/train data/lang exp/tri1 exp/tri1_ali || exit 1;


# Train tri2a, which is deltas + delta-deltas.
steps/train_deltas.sh --cmd "$train_cmd" \
  7500 125000 data/train data/lang exp/tri1_ali exp/tri2a || exit 1;

utils/mkgraph.sh data/lang_test_4g exp/tri2a exp/tri2a/graph_4g || exit 1;

steps/decode.sh --nj 10 --cmd "$decode_cmd" \
  exp/tri2a/graph_4g data/test120_p_spk exp/tri2a/decode_test120_p_spk|| exit 1;


steps/train_lda_mllt.sh --cmd "$train_cmd" \
   --splice-opts "--left-context=5 --right-context=5" \
   7500 125000 data/train data/lang exp/tri1_ali exp/tri2b || exit 1;

utils/mkgraph.sh data/lang_test_4g exp/tri2b exp/tri2b/graph_4g || exit 1;
steps/decode.sh --nj 10 --cmd "$decode_cmd" \
  exp/tri2b/graph_4g data/test120_p_spk exp/tri2b/decode_test120_p_spk || exit 1;


steps/align_si.sh  --nj 10 --cmd "$train_cmd" \
  --use-graphs true data/train data/lang exp/tri2b exp/tri2b_ali  || exit 1;




# From 2b system, train 3b which is LDA + MLLT + SAT.
steps/train_sat.sh --cmd "$train_cmd" \
  7500 125000 data/train data/lang exp/tri2b_ali exp/tri3b || exit 1;

# Trying 4-gram language model
utils/mkgraph.sh data/lang_test_4g exp/tri3b exp/tri3b/graph_4g || exit 1;

steps/decode_fmllr.sh --cmd "$decode_cmd" --nj 10 \
  exp/tri3b/graph_4g data/test120_p_spk exp/tri3b/decode_test120_p_spk || exit 1;

# This is commented out for now as it's not important for the main recipe.
## Train RNN for reranking
#local/sprak_train_rnnlms.sh data/local/dict data/dev/transcripts.uniq data/local/rnnlms/g_c380_d1k_h100_v130k
## Consumes a lot of memory! Do not run in parallel
#local/sprak_run_rnnlms_tri3b.sh data/lang_test_3g data/local/rnnlms/g_c380_d1k_h100_v130k data/test1k exp/tri3b/decode_3g_test1k


# From 3b system
steps/align_fmllr.sh --nj 10 --cmd "$train_cmd" \
  data/train data/lang exp/tri3b exp/tri3b_ali || exit 1;

# From 3b system, train another SAT system (tri4a) with all the si284 data.

steps/train_sat.sh  --cmd "$train_cmd" \
  13000 300000 data/train data/lang exp/tri3b_ali exp/tri4a || exit 1;

utils/mkgraph.sh data/lang_test_4g exp/tri4a exp/tri4a/graph_4g || exit 1;
steps/decode_fmllr.sh --nj 10 --cmd "$decode_cmd" \
   exp/tri4a/graph_4g data/test120_p_spk exp/tri4a/decode_test120_p_spk || exit 1;



# alignment used to train nnets
steps/align_fmllr.sh --nj 10 --cmd "$train_cmd" \
  data/train data/lang exp/tri4a exp/tri4a_ali || exit 1;

## Works
local/sprak_run_nnet_cpu.sh 4g test120_p_spk || exit 1;



# Getting results [see RESULTS file]
for x in exp/*/decode*; do [ -d $x ] && grep WER $x/wer_* | utils/best_wer.sh; done