When you compare Western Music to Indian Classical you will notice that the two are very different when it comes to the fundamental building blocks of music. Indian Classical Music – both the Carnatic and Hindustani forms — are based on Raga along with its varied characteristics.

But what makes up a Raga?

A shruti is the smallest gradation of pitch — it is the smallest gradation of pitch that the human ear can detect. It is also the smallest gradation of pitch that a singer or an instrument can produce. A Swara is  a selected set of pitches from which a musician constructs scales, melodies and ragas.

In other words, Swaras when combined in a specific sequence forms Raga. Each Raga features Swaras Ascending & Descending Sequence namely:

• Arohana
• Avarohana

Both of them can be considered as DNA of Raga. There are various techniques to identify the Raga of music.

Identifying the Basics Swaras

The foremost step for Swara identification is fixing upon “Sa” & “Pa” and further proceed as the base. You should have sound knowledge on the positions of variant varieties of Swaras like “Ri, Ga, Ma, Dha & Ni”.

Additionally, you should have an idea of how these variants are arranged sequentially on the Octave as lower to upper octave as “Sa”. Then similarly proceed towards descending order as upper to lower octave as “Sa”.

It is easy to get insight with instruments including Keyboard or Veena. If you are playing on the violin then also you can feel it easier with fingering techniques used for Swara positions.

Most commonly known as the Parent Raga Melakarta Raga comprises of all the 7 Swaras including:

• S
• R
• G
• M
• P
• D
• N

Janya Raga, the child Raga has been derived from Melakarta Raga (Parent Raga) consisting of only 5 or 6 by omitting 1 or 2 of 7 Swaras. The Child Raga may be thought to have been derived from manifold parent ragas.
Carnatic Music builds a hierarchical relationship for the classification of Raga as a Parent-Child relationship. Here we have taken the help of 3 Approaches to identify both Melakarta Raga and Janya Raga used as Parent-Child in Carnatic Music. In addition to Arohana and Avarohana we should even consider features of Raga.

Apart from Aarohanam and Avarohanam Raga takes into account “Raga Lakshana” expressing semantic details about Raga. It comprises of thirteen important features. Let’s have a look at them:

• Graha: is a note based on which Raga begins
• Amsa: is a note revealing a Raga’s melodic entity (Jiva Swara
• Nyasa: is a note at which Raga may be concluded
• Mandra: is the lowest note to be played in Raga
• Tara: is the highest note to be played in Raga
• Alpatvas are notes used carefully in Raga
• Bahutvas are frequently used notes in Raga
• Apanyasa is similar Sangati being sung in Madhya & Tara sthayi
• Vinayasa: describes the pattern to sing a specific Raga
• Sanyasa is the finally sung Raga, elaborated and concluded at Adhara Shadja Swara
• Shadavas are Sancharas of 6 notes
• Audava are Sancharas of 5 notes
• Antara Marga is a note introduction or another Raga’s Chayya​

There are 2 approaches to identify Raga namely:

• Arohana Avarohana Approach
• Raga Model-Based Approach

We are going to discuss both of them here.

Arohana Avarohana Approach

Swara is the prime component of this Raga consisting of both Arohana and Avarohana. The advantage here is with both the Swaras you can easily identify the Raga. The algorithm in this approach had been carried out with 2 scenarios:

• One using recognized fixed frequency matching up with Shadja “S”
• Other by calculating the tonic with an algorithm to specify Shadja “S” frequency

Songs had been selected of those singers whose tonics are assumed already. Input Signal here is segmented with a segmentation algorithm.
Concluding the segmentation phase, assumption lays on this fact – each segment may likely correspond to Swara. Hence you have identified dominant frequency matching up with each segment.

Dominant Frequency is typically decided using two special features:

• Spectral Energy
• Spectral Centroid

Calculating the frequency: if you find 95% energy is present then it is the Dominant Frequency.

Once you have identified the components of frequency from each component they are converted to notations of Swaras via determining the ratio amid frequency and known tonic. With the help of these Swaras Arohana & Avarohana are decided by selecting 1 from 7 Swaras.

Raga Model-Based Approach

At the phase of Raga Identification, the components of frequency are derived from the signal of input music after extracting characteristics and tonic referring to the “S” frequency.

With such extracted components of frequency and tonic, the ratio amid these components is used to determine the Swaras that make up the input. From these components of Swara, you can identify the Raga Lakshana by monitoring Swaras found in successive segments. The sequences of these Swaras are used to identify the Raga.

Scale Matching

This is another approach you can use to identify the Raga. Based upon the scale it is evaluated with scale accumulated in the database. The raga matching up to the raga in your database is the result of Raga & System’s output. The test Data by “Sridhar” and “Geetha” comprises of thirty samples in three Melakarta Ragas which has been sung by:

• 4 Musicians
• 175 Talam
• Raga Database with the names
• Arohana and Avarohana in component form of Swara

The steps followed under this approach are:

• • Signal Separation – Signal of input music is divided into music & voice
  • Onset Detection – Start of music wherein amplitude increases from “0” to the first peak
  • Offset Detection – Signal amid the onset & offset points are considered
  • First Level Segmentation – Match the results of onset as well as offset detection as the divided signal with the Talam in the database
  • Second Level Segmentation – Segment here is classified in 1, 2 or 4 individual segments of Talam (Swaras) based on tempo within which the song is being sung

Feature Extraction  HPS Algorithm is being used to find out the component of frequency for every Swara. Fundamental Frequency normally corresponding to the Swara “S” is then determined through singer identification.

Feature Mapping & Raga Identification 

Swara corresponding to each frequency is determined by the available ratio, used them, and evaluated with swaras corresponding to a specific Raga in the database.

Students of Carnatic Music spend many months learning these core concepts of Swaras. It requires constant practice and feedback from Gurus during your online or in-person classes. It is also best if students record their practices and share with their Gurus so the Gurus can help the students identify mistakes and correct them so the students sign correctly. Also, when students record their singing and playback, these students can then identify whether they’ve sung the Swaras correctly.

For a no-obligations  trial with one of our esteemed Gurus, please fill out the form below. We can help you schedule a Carnatic Music lesson (Carnatic Vocal, Flute, Vioin, Mridangam). You can also visit our studios here in one of our four location in San Jose, Saratoga, Fremont or Foster City. Qutie a few of our students also attend our online-only lessons.