score_abcd_kaldi.sh

#!/usr/bin/env bash

# usage:
# $ ./score_abcd.sh ./exp/tri3a_dnn/decode_tgpr_5k_dev_0330/scoring/15.tra
# tmp/result.A.txt:       | SPKR   | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
# tmp/result.A.txt:       | Sum/Avg|  330   5353 | 94.2    5.4    0.4    2.5    8.3   52.4 |
# tmp/result.B.txt:      | SPKR   | # Snt  # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
# tmp/result.B.txt:      | Sum/Avg| 1980   32118 | 88.3    9.2    2.5    1.8   13.5   66.3 |
# tmp/result.C.txt:       | SPKR   | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
# tmp/result.C.txt:       | Sum/Avg|  330   5353 | 84.8   13.8    1.4    3.0   18.2   65.2 |
# tmp/result.D.txt:      | SPKR   | # Snt  # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
# tmp/result.D.txt:      | Sum/Avg| 1980   32118 | 73.4   20.1    6.4    2.5   29.1   76.2 |

hyp=$1
dir=$(dirname $hyp)/sclite
hyptxt=$dir/$(basename $hyp).txt
reftxt=$(dirname $hyp)/test_filt.txt
mkdir -p $dir

# to text
cat $hyp | utils/int2sym.pl -f 2- ./exp/tri3a_dnn/graph_tgpr_5k/words.txt | sed 's:<UNK>::g' > $hyptxt

# to sclite format
python3 <<EOF
def f(src, dst):
    fd = {k: open(f"{dst}.{k}.trn", "w") for k in ["A", "B", "C", "D"]}

    with open(src, "r") as f:
        for line in f:
            xs = line.split()
            uid = xs[0]
            txt = " ".join(xs[1:]).strip()
            if uid[-1] == "0":
                s = "A"
            elif uid[-1] in "123456":
                s = "B"
            elif uid[-1] == "7":
                s = "C"
            elif uid[-1] in "89abcd":
                s = "D"
            else:
                assert False, f"invalid uid: {uid}"
            fd[s].write(f"{txt}({uid[:4]}_{uid[4:]})\n")

f("$hyptxt", "$dir/hyp")
f("$reftxt", "$dir/ref")
EOF

sclite -r ${dir}/ref.A.trn trn -h ${dir}/hyp.A.trn trn -i rm -o all stdout > ${dir}/result.A.txt
sclite -r ${dir}/ref.B.trn trn -h ${dir}/hyp.B.trn trn -i rm -o all stdout > ${dir}/result.B.txt
sclite -r ${dir}/ref.C.trn trn -h ${dir}/hyp.C.trn trn -i rm -o all stdout > ${dir}/result.C.txt
sclite -r ${dir}/ref.D.trn trn -h ${dir}/hyp.D.trn trn -i rm -o all stdout > ${dir}/result.D.txt

grep -e Avg -e SPKR -m 2 ${dir}/result.*.txt