diff options
| -rw-r--r-- | .travis.yml | 3 | ||||
| -rwxr-xr-x | ship/bin/punani | 1 | ||||
| -rwxr-xr-x | ship/build | 104 | ||||
| -rwxr-xr-x | util/bin/mic.stt | 12 | ||||
| -rw-r--r-- | util/lib/stt/README.md | 4 | ||||
| -rw-r--r-- | util/lib/stt/google.sh | 35 | ||||
| -rwxr-xr-x | util/t/stt/stt-works-with-espeak | 16 |
7 files changed, 142 insertions, 33 deletions
diff --git a/.travis.yml b/.travis.yml index 12f6de4d..1218d9d9 100644 --- a/.travis.yml +++ b/.travis.yml @@ -8,8 +8,9 @@ notifications: script: "! ( make -C util test | grep '^not ok' )" before_install: - sudo apt-get install bc -qq - - sudo apt-get install python -qq - sudo apt-get install w3m -qq +# stt-espeak test + - sudo apt-get install flac espeak -qq branches: only: - master diff --git a/ship/bin/punani b/ship/bin/punani index 99a5a813..28bf7d1a 100755 --- a/ship/bin/punani +++ b/ship/bin/punani @@ -1,3 +1,4 @@ #!/bin/sh +#@info #@include punani punani "$@" @@ -2,67 +2,107 @@ set -euf ## SYNOPSIS -# build compile SRCFILE DSTFILE -# build deps SRCFILE +# [debug=true] build compile SRCFILE DSTFILE +# [debug=true] build deps SRCFILE... build() { case "$1" in compile) build_compile "$2" "$3";; - deps) build_deps "$2";; + deps) shift; build_deps "$@";; *) echo "build: $1: bad command" >&2; return 23;; esac } +# usage: debug_script VARNAME [DESCRIPTION] +debug_script() { + if test "${debug-false}" = true; then + printf '[35m====== %s%s[m\n%s\n' \ + "$1" \ + "${2+" ($2)"}" \ + "$(eval echo \"\$$1\" | nl -b a)" >&2 + fi +} + ## build directives +build_info_directive='#@info' build_include_directive='#@include \([0-9A-Za-z]\+\)' +input_parser="\ +s:^ *\([0-9]\+\) "$build_info_directive"$:build_info \1: +s:^ *\([0-9]\+\) "$build_include_directive"$:build_include \1 \2: +t +s:^ *\([0-9]\+\) .*:echo \1p:" +debug_script input_parser + +## usage: build_include LINENO LIBNAME +build_include() { cat<<EOF +$1a\\ +# begin $2 +$1r$(build_resolve $2) +$1a\\ +# end $2 +EOF +} + +## usage: build_info LINENO +build_info() { cat<<EOF +$1a\\ +# this file was generated by //ship/build\\ +# date: $(date -u --rfc-3339=s)\\ +# version: $(git rev-parse HEAD) +EOF +} + ## usage: build_compile SRCFILE DSTFILE build_compile() { - cp "$1" "$2" + srcfile="$(cat "$1")" + debug_script srcfile 'SRCFILE' - while needs_compilation "$2"; do - script="$(make_sedscript_maker_shellscript "$2" | sh)" - sed -n "$script" "$2" >"$2.tmp" - mv "$2.tmp" "$2" + while needs_compilation "$srcfile"; do + script="$(make_sedscript_maker_shellscript "$srcfile")" + srcfile="$(echo "$srcfile" | sed -n "$script")" + debug_script srcfile 'sed sedscript srcfile' done + echo "$srcfile" > "$2" chmod +x "$2" } -## usage: needs_compilation SRCFILE +## usage: needs_compilation SHELLSCRIPT # Returns true if SRCFILE contains compilation directives. needs_compilation() { - grep -q "^$build_include_directive$" "$1" + echo "$1" | + grep -q "^\\($build_include_directive\\|$build_info_directive\\)$" } ## usage: make_sedscript_maker_shellscript SRCFILE # Print a shellscript that creates a sedscript that resolves all the build # directives in SRCFILE. make_sedscript_maker_shellscript() { - echo 'set -euf' - deps="$(build_deps "$1")" - for lib in $deps; do - echo "_build_include_$(basename $lib)=$lib" - done - nl -b a -s ' ' "$1" | - sed ' - s:^ *:: - s:^\([0-9]\+\) '"$build_include_directive"'$:\1a\\\\\ -# BEGIN \2\ - \1r\$_build_include_\2\ - \1a\\\\\ -# END \2: - s:^\([0-9]\+\) .*:\1p: - - 1s:^:echo ": - $s:$:": - ' + sedscript_generator="$(echo "$1" | nl -b a -s ' ' | sed "$input_parser")" + debug_script sedscript_generator 'sed input_parser srcfile' + + sedscript="$(eval "$sedscript_generator")" + debug_script sedscript 'eval sedscript_generator' + + echo "$sedscript" } -## usage: build_deps SRCFILE +## usage: build_deps SRCFILE... +# Print all the dependencies of SRCFILE... (in alphabetic order) to stdout. build_deps() { - for libname in $(sed -n 's:^'"$build_include_directive"'$:\1:p' "$1"); do - build_resolve "$libname" - done + while test $# -gt 0; do + deps="$( + for f; do + for d in $(sed -n 's:^'"$build_include_directive"'$:\1:p' "$f"); do + build_resolve $d + done + done + )" + set -- $deps + if test $# -gt 0; then + echo "$deps" + fi + done | sort | uniq } ## usage: build_resolve LIBNAME diff --git a/util/bin/mic.stt b/util/bin/mic.stt new file mode 100755 index 00000000..9236f85c --- /dev/null +++ b/util/bin/mic.stt @@ -0,0 +1,12 @@ +#!/bin/sh +set -efux +cd $(dirname $(readlink -f $0)) +. ../lib/stt/google.sh +duration=${1?please provide duration via \$1} +lang=${lang:-de-DE} +export lang +echo "language is set to $lang" +echo "will record for '$duration' seconds" +f=$(record_audio ${duration}) +trap 'rm $f' TERM EXIT HUP +stt "$f" diff --git a/util/lib/stt/README.md b/util/lib/stt/README.md new file mode 100644 index 00000000..be905770 --- /dev/null +++ b/util/lib/stt/README.md @@ -0,0 +1,4 @@ +# Speech to Text api wrapper + +Because Speech to text is hard™ with FOSS, these libraries utilize the magic of +the internets to solve this problem. diff --git a/util/lib/stt/google.sh b/util/lib/stt/google.sh new file mode 100644 index 00000000..a78579d5 --- /dev/null +++ b/util/lib/stt/google.sh @@ -0,0 +1,35 @@ +#!/bin/sh + +_get_content_type(){ + file -b --mime-type "$1" +} +_get_audio_rate(){ + file "$1" | sed -n -e 's/.* \([.0-9]\+\) kHz.*/\1/p' \ + | awk '{print int($1 *1000)}' +} + +record_audio(){ + # usage : _record_audio num_seconds + # echoes the output file + tmpfile=$(mktemp) + : ${1?please provide number of seconds to record} + arecord -d "$1" -r 16000 -t wav -q -f cd | flac -s -f - -o "$tmpfile" && echo "$tmpfile" +} +stt(){ + # usage: (lang=de-de stty recorded_file) + : ${1? please provide recorded file} + infile="$1" + lang=${lang:-en-us} + _get_content_type "$1" | (! grep -q "x-flac" ) \ + && echo "infile needs to be in flac format" \ + && return 1 + # only flac seems to be working... + wget -q -O - \ + -U 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_6_8) AppleWebKit/535.7 (KHTML, like Gecko) Chrome/16.0.912.77 Safari/535.7' \ + --post-file "$infile" \ + --header "Content-Type: `_get_content_type $infile`; rate=`_get_audio_rate $infile`;" \ + "http://www.google.com/speech-api/v1/recognize?lang=${lang}&client=chromium&maxresults=1" \ + | sed -n 's/.*utterance":"\([^"]*\)".*/\1/p' + + # returns {"status":0,"id":"d9269e6f741997161e41a4d441b34ba1-1","hypotheses":[{"utterance":"hallo Welt","confidence":0.7008959}]} +} diff --git a/util/t/stt/stt-works-with-espeak b/util/t/stt/stt-works-with-espeak new file mode 100755 index 00000000..ff39f567 --- /dev/null +++ b/util/t/stt/stt-works-with-espeak @@ -0,0 +1,16 @@ +#!/bin/sh + +cd $(dirname $(readlink -f $0)) +. ../../lib/stt/google.sh +tmp=$(mktemp) +test_str="hello" +trap "rm $tmp" TERM INT EXIT HUP +espeak --stdout "$test_str" | flac --totally-silent -f -o "$tmp" - + +if stt "$tmp" | egrep "^$test_str\$" >/dev/null ;then + echo "ok" + exit 0 +else + echo "not ok" + exit 1 +fi |