Creative

What Instrument Is This? Identify Instruments by Sound

Upload a song or clip and find out which instruments you are hearing and what each one is doing.

Instrument Detector

You hear a haunting reed sound in a film score, or a plucked tone in a song that is not quite a guitar, and you want to know what it is. Upload the recording and the AI lists every instrument it can pick out, each with a confidence score, so you can see which answers are firm and which are educated guesses. It goes beyond naming: it explains whether each instrument carries the lead, the rhythm, the bass line or the percussion, whether it is acoustic or electronic, which playing techniques are in use, how prominent it sits in the mix, and where the arrangement shifts between solo passages and the full ensemble. Uncommon or regional instruments, such as a duduk, a hang drum or an erhu, are called out specifically.

Example result

Input

A 45-second clip from a film soundtrack with a breathy wind melody over low strings and soft percussion.

Output

Instruments detected: Duduk (lead): the breathy, nasal double-reed tone and narrow range point strongly to an Armenian duduk rather than an oboe. Played legato with slow bends between notes. Cello section (harmony and bass): sustained bowed chords underneath; the lowest line may be doubled by double bass. Frame drum (percussion): soft hand strikes on the downbeats, acoustic. Synth pad (texture): a faint electronic swell enters at 0:28, low in the mix. Arrangement: the duduk plays alone for the first eight seconds, then strings join; the ensemble thickens after 0:28. Prominence: the duduk sits well forward; percussion is mixed quietly. Rare instrument noted: duduk, an apricot-wood reed instrument common in Armenian music and film scores.

Scores

Duduk86
Cello91
Double bass58
Frame drum72
Synth pad64

How it works

How Instrument Detector works

  1. 01

    Share the passage

    Sign in, open the tool and upload the song or a clip that contains the sound, or record it through your browser from a nearby speaker. Choose the AI model to use.

  2. 02

    Timbre and register

    The AI listens to the attack, sustain and tone color of each voice in the mix, along with its pitch range, to tell a cello from a viola, a Rhodes from a Wurlitzer, or a real kit from a drum machine.

  3. 03

    Roles, layers and technique

    It works out which part each instrument plays, how the layers stack over time, and how it is played: bowed or pizzicato, fingerpicked or strummed, muted trumpet or open, slide or fretted.

  4. 04

    Confidence and rarities

    Each instrument gets a 0 to 100 confidence score. Rare or culturally specific instruments are flagged with a short note on what makes them recognizable.

Use cases

Who it is for

Curious listeners

Settle the question of what that sound in a song, ad or soundtrack actually is, and learn the name so you can find more music that features it.

Musicians transcribing or covering

Find out every part in an arrangement before you learn it, including quiet layers like a doubled bass line or a tucked-away tambourine.

Producers matching a reference

Identify whether a reference track uses a real piano, an electric piano or a synth patch, and whether the drums are acoustic or programmed, before choosing your own sounds.

Teachers and ensemble directors

Turn an orchestral or world-music recording into a listening lesson, with the report as the answer key for which sections play when.

Tips

Get better results

  • Trim the upload so it includes the moment where the mystery instrument is loudest or plays alone.
  • Use the original file or a streaming-quality download; a phone held to a laptop speaker loses the high overtones that separate similar instruments.
  • If the instrument appears only briefly, mention the timestamp in your own notes and check that part of the report first.
  • Pay attention to the confidence score: below about 60 usually means the AI is choosing between close relatives, so treat the name as a lead to verify.
  • For dense mixes, try a second upload of a quieter section; fewer layers means cleaner identification.

Limitations

What it can't tell you

Identification is based on what the recording sounds like, so heavily processed or sampled sounds can be mistaken for the instrument they imitate, and a good synth emulation of strings may be reported as strings. Closely related instruments (viola and violin, alto and tenor saxophone, various lutes) are the hardest to separate. The report cannot tell you the brand or model of an instrument, and confidence scores are the AI's own estimate, not a measured probability.

Frequently Asked Questions

Instrument Detector questions

How do I identify an instrument by sound?

Upload a clip where the instrument is clearly audible. The AI compares its tone color, attack, range and playing style with the instruments it knows and returns a name plus a confidence score. If the sound only lasts a moment, trimming the clip around it gives the clearest answer.

Can it identify several instruments at once?

Yes. It lists every instrument it can separate in the track, from the lead melody down to quiet background percussion, and describes each one's role and how loud it is in the mix. Very dense productions may hide the quietest layers.

Can it tell real instruments from electronic ones?

It reports whether each sound seems acoustic or electronic, for example a live drum kit versus a drum machine, or an acoustic piano versus a digital one. High-quality sample libraries can be convincing, so treat that label as an informed judgment.

Does it recognize folk and world instruments?

It is asked specifically to point out unique or rare instruments, so sounds like the sitar, oud, koto, kora, bagpipes or steel pan are named when they are present, with a short explanation of the features that gave them away.

Is this the same as song identification?

No. This tool tells you which instruments are playing, not the title or artist of the song. To find out what a song is called, use the song identifier instead.

Instrument Detector uses credits from a paid LindaleAI plan. See pricing