Free Japanese speech to text: live transcription

Speak Japanese, polite or casual, Tokyo or Osaka, and watch your speech turn to text in kanji and kana as you talk, free and with no sign-up.

クリックまたはSpaceで文字起こしを開始
タップして文字起こしを開始
話すテーマに迷ったら
今週末は何をする予定ですか?

Join the teams making the switch to Cartesia

Artificial Analysis

Ranked #1

in Speech Arena leaderboard & Speech to Text leaderboard by Artificial Analysis

How to convert Japanese speech to text

Have a recording instead? The audio to text converter takes Japanese files and detects the language on its own.

01

Press the microphone button and allow access when your browser asks.

02

Speak Japanese the way you normally would. Each word appears as you say it.

03

Select the transcript, copy it, and paste it into your document, email, or notes.

Add Japanese speech to text to your app

Open a WebSocket to Ink-2 and stream Japanese audio in. Ink-2 detects the language on its own, so there is nothing to set. Transcripts and turn events come back as the person talks. The speech-to-text docs list every supported language.

Model
ink-2
Languages
enesfrhija
Turn events
turn.startturn.updateturn.end
transcribe.ts
import { readFile } from "node:fs/promises";
import { Cartesia } from "@cartesia/cartesia-js";

const client = new Cartesia({ apiKey: process.env.CARTESIA_API_KEY });

const ws = await client.stt.autoFinalize.websocket({
  model: "ink-2",
  encoding: "pcm_s16le",
  sample_rate: 16000,
});

ws.on("turn.update", (event) => console.log(event.transcript));
ws.on("turn.end", (event) => console.log("final:", event.transcript));

// 16-bit PCM, 16 kHz, mono. Send it in 100 ms chunks.
const audio = await readFile("speech.pcm");
for (let offset = 0; offset < audio.length; offset += 3200) {
  const chunk = audio.subarray(offset, offset + 3200);
  ws.sendRaw(chunk);
}

Japanese speech to text FAQs

Is there a free Japanese voice to text tool?

Yes. The tool on this page turns your Japanese speech into text for free, with no account, up to a daily limit. A free Cartesia account's monthly credits cover up to 111 minutes of transcription through the API, and paid plans add more.

Can I transcribe Japanese audio files?

Yes. Upload an MP3, WAV, M4A, WebM, FLAC, or OGG file up to 10 minutes and 200 MB to the audio to text converter. It detects Japanese on its own, so there is nothing to set before you upload. No account needed.

Does it handle Kansai and other dialects?

There is nothing to pick. Ink-2 has no setting for a dialect, and it detects Japanese on its own. Speak into the mic in your own dialect and read the transcript.

How are Japanese phone numbers read aloud?

Digit by digit, in the groups they are written in, with "no" (の) between groups. Zero is "zero" or "rei", so 090-1234-5678 is "zero kyū zero no ichi ni san yon no go roku nana hachi".

How do I turn Japanese audio to text with the API?

Open a WebSocket to Ink-2 and stream the audio in. There is no language to set: Ink-2 detects the language on its own, so you open the connection with the model, the audio encoding and the sample rate, and stream Japanese audio in. The speech-to-text docs list Japanese (ja) among the five languages Ink-2 supports.

Can Cartesia read Japanese text aloud as well?

Yes. Cartesia's text-to-speech model covers 44 languages, Japanese among them. Try it on the Japanese text to speech page.