whisper-cppによる文字起こしスクリプト

whisper-transribe.sh

#!/bin/bash
set -euo pipefail

WHISPER_MODELS="${WHISPER_MODELS:-/Volumes/YourDrive/whisper-models}"

usage() {
  cat <<EOF
Usage: $(basename "$0") [-l lang] [-m model] <input_file>

Options:
  -l lang    Language code (default: ja)
  -m model   Model name: tiny, base, small, medium, large-v3, etc. (default: medium)

Environment:
  WHISPER_MODELS  Directory containing ggml model files
                  (default: /Volumes/YourDrive/whisper-models)

Examples:
  $(basename "$0") recording.m4a
  $(basename "$0") -l en -m large-v3 meeting.mp4
EOF
  exit 1
}

lang="ja"
model="medium"

while getopts "l:m:h" opt; do
  case "$opt" in
    l) lang="$OPTARG" ;;
    m) model="$OPTARG" ;;
    h) usage ;;
    *) usage ;;
  esac
done
shift $((OPTIND - 1))

[[ $# -lt 1 ]] && usage

input="$1"
model_file="${WHISPER_MODELS}/ggml-${model}.bin"

if [[ ! -f "$input" ]]; then
  echo "Error: input file not found: $input" >&2
  exit 1
fi

if [[ ! -f "$model_file" ]]; then
  echo "Error: model file not found: $model_file" >&2
  echo "Run: whisper-cpp-download-ggml-model $model $WHISPER_MODELS" >&2
  exit 1
fi

for cmd in ffmpeg whisper-cli; do
  if ! command -v "$cmd" &>/dev/null; then
    echo "Error: $cmd is not installed" >&2
    exit 1
  fi
done

timestamp=$(date +%Y%m%d_%H%M%S)
random_suffix=$((RANDOM % 10000))
wav_tmp="/tmp/whisper-${timestamp}-${random_suffix}.wav"

cleanup() {
  rm -f "$wav_tmp"
}
trap cleanup EXIT

ffmpeg -i "$input" -ar 16000 -ac 1 -c:a pcm_s16le "$wav_tmp" -y -loglevel error

whisper-cli -m "$model_file" -l "$lang" -f "$wav_tmp"