無料オンライン音声テキスト変換
AIによる文字起こしで音声をテキストに変換。音声ファイルをアップロード、マイクで録音、またはURLを貼り付け。100以上の言語、10以上のモデル、98%以上の精度。
1. 音声をアップロード
MP3、WAV、M4A、FLAC、OGGなど、あらゆる音声形式に対応。最大2GB。
2. AIが音声を処理
AIが音声から話者検出とタイムスタンプ付きで音声を抽出。
3. 文字起こしを取得
閲覧、編集、ダウンロード、共有。TXT、SRT、VTT、DOCX、PDFでエクスポート。
音声テキスト変換モデル
ニーズに合ったAIモデルを選択 — または最適なモデルをお任せください。
100以上の言語で音声を文字起こし
音声をテキストに変換する準備はできましたか？
無料で始める →よくある質問
Upload your audio file or paste a URL, pick an AI model, and click Transcribe. STT.ai returns editable text with timestamps and speaker labels — most files finish in under five minutes.
MP3, WAV, M4A, FLAC, OGG, AAC, AMR, and 10+ more are all supported. You don't need to convert between formats first — upload whatever your recorder or app produces.
A little. Lossless formats like WAV and FLAC carry bit-perfect audio, so accuracy is bounded only by the model and speaker clarity. Lossy formats (MP3, M4A) at 128 kbps or higher are effectively identical; very low bitrates under 64 kbps can cost a few points.
Yes. STT.ai includes 600 free minutes per month with no signup for your first file. Paid plans starting at $5/month add longer files, private transcripts, and priority processing.
On clean audio our best models reach 95-97% accuracy (3-5% Word Error Rate). Background noise, overlapping speakers, and strong accents are the main factors that lower accuracy.
Yes. Free users can transcribe up to one hour per file; paid plans extend that to 8+ hours, which covers full-length podcasts, interviews, and audiobooks in a single pass.
Yes. Speaker diarization labels each voice (Speaker 1, Speaker 2, ...) and you can rename them in the editor — works on every supported audio format and model.
Export to TXT, DOCX, PDF, JSON, or SRT/VTT subtitles. JSON keeps machine-readable timestamps and speaker labels; DOCX and PDF are best for sharing and archiving.
Yes. 100+ languages with auto-detection, plus the option to set the language manually. Mixed-language audio is handled by switching mid-file, and you can translate the result afterwards.
Yes. Audio is processed and deleted by default, and Pro plans add client-side encryption so transcripts are unreadable without your key. Nothing is used for training without explicit opt-in.
Yes. Paste a link from any of 1,300+ supported platforms — podcast hosts, SoundCloud, YouTube, and more — and STT.ai fetches the audio directly. DRM-protected sources can't be transcribed.
Yes. The REST API accepts audio files directly, with Python and Node.js SDKs and a free tier of 100 minutes/month. Per-second billing applies beyond the free tier.