sonicAPI.com -- audio processing web service

analyze/melody

Extract the (MIDI) notes from a single voiced audio file.

The melody analysis task detects the note pitches in a monophonic audio file, such as a recording of a vocal or single-voiced instrument. It can also return a detailed pitch contour curve. The result can be used to generate a score-like representation of the audio or to convert audio to MIDI. Furthermore, the results can be re-used in the elastiqueTune task to modify individual notes. For more details on changing individual notes, see process/elastiqueTune.

Parameters

Parameter	Type	Description
detailed_result optional	bool [false, true] default: false	Include additional data (raw pitch curve) in the task result. more This can result in a very long list of pitch marks in your analysis result, so you should only turn this on if you really need it.

Parameter	Type	Description
access_id required	string	One of your personal access_ids.
input_file required	file	A file id or file url or a direct file upload (requires POST). more Upload a local file by performing a POST request with encoding-type "multipart/form-data" (corresponding to an html form with an <input type="file" />)
begin_seconds optional	float	Skip audio data before this position.
end_seconds optional	float	Ignore audio data after this position.
blocking optional	bool [true, false] default: true	Whether to call the task in a blocking or non-blocking way. more Blocking calls return the results directly. For more info see the response documentation for this task.
format optional	string [xml, json, jsonp, xmlp] default: xml	The format to use in the response. more Note that for the formats "jsonp" or "xmlp", the HTTP status code will always be 200.
callback optional	string	Callback function name required when using the format "jsonp" or "xmlp."
foo_string optional	string	Arbitrary string to add to the logs. more Allows you to add any strings you might find helpful for later analysis of the request logs.

Response

blocking=false

The response gives you a status code, the file_id and the corresponding download URL of the resulting file (audio for process tasks and xml for analyze tasks). For processing reports, use the /file/status request with the parameter format=xml/json/jsonp/xmlp.

Name	Description
status	The status code of the task.
file_id	The unique identifier of the file.
href	The direct download link to the file including the file_id.

blocking=true

The results contains the estimated key and frequency of the input file, a list of all detected notes and optionally the raw pitch curve (if the detailed_result parameter was set to true).

Name	Type	Description
key	string	The name of the extracted key. more The algorithm is able to detect both major (Maj) and minor (min) scales from 12 different root notes (C, C#, D, D#, E, F, F#, G, G#, A, A#, B). There are no flat root notes as we assume enharmonic equivalency (c sharp equals d flat).
key_index	integer	A representation of the key as an index. more The key_index can take the values 0...23. The first 12 indices indicate the major keys, and remaining indices the minor keys, both starting from C (example: d# min is 15).
tuning_frequency	float	The melody's tuning frequency in Hz. more The tuning frequency is the frequency of the concert pitch A4 and frequently equals 440Hz. It may deviate a few Hertz for some songs.
notes		List of notes extracted from the music. more Each note has several attributes with information on the position, duration and pitch (see below).

Each note has the following attributes:

Name	Type	Description
midi_pitch	float	The average pitch (perceived frequency) of the note as MIDI pitch index. more Example: C4 (Middle C) is represented by the midi_pitch 60.0; a midi_pitch 60.5 would be 50 Cents higher. All MIDI pitches are given with respect to a tuning frequency of 440 Hz. If, during a note, a frequency deviation such as vibrato is present, then the average pitch will be the centre frequency.
onset_time	float	The beginning time of the note (in seconds)
duration	float	The duration of the note (in seconds)
volume	float	The volume or velocity of the note (normalized to the range between 0 and 1)

Each pitch_mark in the optional pitch_curve result has these attributes:

Name	Type	Description
tonal	string	"true" for tonal sections, "false" if the signal is either silent or non-tonal.
midi_pitch	float	The instantaneous pitch at the pitch mark position as MIDI pitch index.
time	float	The position of the pitch mark (in seconds)
volume	float	The instantaneous volume at the pitch mark position (normalized to the range between 0 and 1)

One-click example

You can analyze the melody of this saxophone recording with a simple click on this button:
Analyze Learn more The button is just a link with a specially constructed URL:
https://api.sonicapi.com/analyze/melody
?access_id=584ee88b-4525-4f14-b2ea-94dedda17a30
&input_file=http://www.sonicapi.com/music/solo_sax.mp3 By requesting this URL, the input file will be imported into the system, processed by sonicAPI and the analysis result is displayed in your browser. The Live-Demo generates additional example code.