v. to mutter · to murmur · to mumble to oneself · (modern slang) to tweet
IPA tsɯ·bɯ·ja·kɯ̥accent LHHH (平板 heiban, no internal drop)4 mora · all open
呟
Why the study card sounds different from the song
Textbook audio is syllable-timed — every mora gets equal weight, no devoicing, neutral pitch accent. Singing is none of that. The two tsu/bu opens get soft and almost identical, the final ku usually devoices into a near-silent k(u), and the melody overwrites the pitch accent entirely.
Trying to learn the word from a flashcard and then meeting it in a song is like trying to learn Billy Joel from Billie Joe Armstrong's cover. Same notes, different mouth. Below the same word performed in four registers — listen top to bottom and you'll hear the drift.
The drift — same word, four registers
1 · card
Study-card clean. Every mora separate. This is what the flashcard plays.Google Chirp 3 HD · Kore (neutral)
2 · warm
Narrator-warm. Real pacing, real intonation. Closer to spoken Japanese.Google Chirp 3 HD · Charon (warm narrator)
3 · muttered
The word performing its meaning. Vowels softened, final ku half-swallowed.Qwen3-TTS instruct「つぶやくように、独り言のように、小さな声で」
4 · sung
Sung style. Vowels stretched. Pitch accent overwritten by the melody.Qwen3-TTS instruct「歌うように、ゆっくり、メロディに乗せて」
The kanji is a memory trick
呟
口 + 玄
a mouth + something dark / hidden / mysterious
The picture is the meaning. The 口 is a mouth. The 玄 element means "dark, profound, hidden" (same character in 玄関, the entry threshold). A mouth speaking something private, half-hidden — that's muttering. The word is built for it.
The sound is the act.tsu-bu has the shape of a small grain or drop (粒, also tsubu) — tiny words, dropping out of your mouth one at a time. -yaku is the verb-tail. You're producing tiny utterances.
Pitch accent. 呟く is heiban (flat, type 0) — no drop inside the word. All four morae climb low-to-high and stay there. If you put a stress in the middle you're singing the wrong song.
Final-u devoicing. Tokyo speakers routinely devoice the final ku → k(u), voiceless, almost lost. Whispered/muttered registers devoice it even more. The study card preserves it. Real speech rarely does.
つ
ぶ
や
く
LOWHIGHHIGHHIGH
drop occurs on the following particle, not inside the word
Don't mix it up with ささやく
These two get translated the same in English ("whisper") but they're different acts in Japanese. If you pick the wrong one you sound off.
つぶやく
tsubuyaku · 呟く
Solitary. Internal. Lonely. You mutter to yourself, into the void, under your breath. This is the verb Twitter Japan used for "to tweet." Lyrics use it for inward thoughts and quiet laments.
ささやく
sasayaku · 囁く
Intimate. Directed. Romantic. You whisper to someone, close to the ear. Lyrics use it for sweet nothings, the wind speaking, secrets shared.
The noun form is the one in lyrics
Songs often use the noun form つぶやき (tsubuyaki, "a mutter / a murmur") more than the verb. Same root, same kanji 呟, conjugated differently. Hear them side by side:
In context — phrases you'll meet
Common collocations. Warm-narrator voice — closer to what your brain has to recognize in a song than the study card.
Voice playground — same line, eight performances
Same text 「つぶやく。心の中でつぶやく。君の名前をつぶやく。」 rendered eight ways. Notice what each engine drops, stretches, or emphasizes — and whether the muttered/sung instructs actually land.
How to practice
A pattern that works for sticky Japanese words, in order:
Hear it three different ways before you try to say it. Card → warm → muttered, in that order. Each one teaches a different layer.
Mutter it yourself three times before you say it at full volume. Muttering forces the right register — low, soft, slightly slurred — that the textbook voice scrubs out.
Slow-sing the lyric line, holding each vowel longer than feels natural, then snap back to song tempo. Slow pass locks the mouth shapes; snap-back keeps them at speed.
Tune to the warm narrator (Charon above), not the study card. The card voice is for memorizing characters. The warm narrator is closer to the song.