How to use Chandas.org

Just start typing in a Kannada, Telugu, Gujarati, or Devanagari verse—or choose a Roman input scheme above the slate. For English verse, explicitly choose English · stress meter. Chandas.org marks guru laghu and counts the syllables, so you can save your arithmetic for less poetic occasions. Looking to learn various Chandassu/meters? Jump directly to the supported-meter field guide.

Write in Roman transliteration

Choose IAST, ISO 15919, ITRANS, or Harvard-Kyoto in the Input control. Chandas quietly makes an offline Devanagari shadow for metrical arithmetic, then puts every underline back on the Roman letters you typed. Your poem does not visit a transliteration server. ISO and IAST are separate choices because their vowel conventions are genuinely different, not because the menu wanted more exercise.

For reversible southern-script IAST, short e/o appear as ĕ/ŏ, and retroflex ಳ appears as l̤. This keeps it distinct from vocalic ḷ so a trip to Roman and back does not quietly replace a consonant with a vowel.

Roman mode supports Ghost templates with L/G guidance. Strong templates are temporarily disabled in Roman mode; switching from a Strong native-script draft safely returns it to the ordinary slate. The chosen scheme is retained in saved poems, backups, and analysis links.

The same Input menu also contains Convert entire composition to. It previews a whole-buffer conversion to Kannada, Devanagari, Telugu, Gujarati, IAST, ISO 15919, ITRANS, or Harvard-Kyoto before replacing anything, and offers Undo afterward. Meter selections and verse layout stay put. Colloquial Roman is a copy-only preview: it cheerfully drops distinctions that the meter checker cannot afford to forget.

Write in English stress meter

Choose English · stress meter in the Input menu. Nothing in the ordinary Indic or Roman-transliteration composer changes until you do. Chandas then marks weak syllables with a dotted green underline and strong syllables with a solid saffron underline, draws faint foot boundaries, and reports syllables and stresses from the beginning of the current line to the cursor.

Select an English meter per stanza to see departures in red and use its Ghost guide. Strong guidance keeps whole words or phrases in metrical cells; it never chops spelling into pretend syllables. Dashed uncertainty marks mean the pronunciation or spelling-to-syllable alignment was inferred; they are a request for human judgment, not a duel challenge from the dictionary. In the stress pattern, || marks the first beat—the tāḷa entry—and anything before it is the pickup. Surplus opening syllables outside the meter are marked as extrametrical anacrusis. English may promote or demote stress to fit the spoken rhythm. Open Adjust stress & pronunciation to give any word the reading you actually intend; poets retain veto power. English mode, selected meters, and Ghost guidance survive local recovery and analysis links.

Pickup, anacrusis, and the first beat

These terms overlap, but they are not interchangeable. Chandas uses them this way:

Pickup

The performance-friendly label for syllables shown before ||. An expected opening weak position may be a pickup without being outside the meter.

Anacrusis

One or more normally unstressed opening syllables before the verse proper. When they lie outside a fixed template, Chandas calls them extrametrical anacrusis, also described in scholarship as initial extrametricality or Auftakt.

Ictus

A strong metrical position or metrical accent. Chandas uses || for the chosen first ictus—the entry of the tāḷa. It is analogous to a musical downbeat, not literally a notated bar line.

The rare adjective procephalic may describe a line with an extra opening syllable, but Chandas avoids it as a general label. Hypercatalectic concerns extra material at the end of a metrical line, so “hypercatalectic in reverse” is not our term for anacrusis. Nor is “epicapital,” which is not an established prosodic term. In short: an anacrusis is itself a kind of pickup; it does not come “before the first pickup.” See the Allen and Greenough prosody glossary and the Cambridge account of anacrusis and Auftakt.

The compact pronunciation pack loads only after English is first selected and is cached for later offline use. A second compact pack then compares line-ending pronunciations and shows an end-rhyme scheme such as A B A B. Chandas recognizes perfect rhyme from the final stressed vowel onward, labels masculine and feminine endings, and leaves unknown words as ? rather than inventing a rhyme. All analysis remains on the device. Switching back to native or Roman input restores the existing Guru–Laghu interface.

When enough lines agree, a separate advisory panel may suggest a poetic form: blank verse, heroic couplet, common measure, ballad stanza, long or short measure, fourteener verse, Poulter's measure, limerick, common limerick, English/Spenserian/Petrarchan sonnet, rhyme royal, ottava rima, terza rima, or Spenserian stanza. Forms combine line count, line meter, and rhyme; they do not crowd the meter picker or turn themselves into red violations. Slant, eye, and internal rhyme are labeled as evidence, separately from dictionary-perfect rhyme.

M6/M7 add two-to-five-beat sprung rhythm, where stress-led feet can sit next to each other and carry variable slack, plus modern alliterative verse. For the latter, write || at the caesura when you know it; Chandas compares the stressed onsets across the two half-lines. This is a modern composition guide, not a claim to classify Old English or Sievers types.

Open English ear tools for a North American dictionary reading, cautious non-rhotic rhyme hints, or experimental Early Modern eye-rhyme hints. The same drawer can learn the current stanza as your own named English form. Learned forms stay on the device and travel in a full backup. Historical hints are clues, not a time machine with excellent diction.

A strict Limerick follows the familiar anapestic long/long/short/short/long frame. Common limerick keeps five lines, perfect AABBA rhyme, and the 3 / 3 / 2 / 2 / 3 beat contour, but accepts compatible iambic, trochaic, or anapestic realizations. If the strict form fits, Chandas shows only “Limerick”; the common form is its flexible fallback, not an eager duplicate.

If five lines have a fully known AABBA rhyme scheme Chandas also calls them Limerick-y. A stricter Limerick or Common limerick may appear above it; Limerick-y makes a modest additional claim about the rhyme and line shape—not a suspiciously confident claim about the rhythm.

Read the markings

A dotted underline marks Laghu; a solid underline marks Guru. Red means the selected meter disagrees (not that the app has formed an opinion about the poet).

Kannada, Telugu, Gujarati, and Devanagari are analyzed directly on this device—no transliteration service receives the poem. Script support applies the shared syllable-length catalog; it does not quietly claim that every language's native meter tradition is the same. Gujarati also has its own sourced catalog for Caupāī, Doharō, Soraṭhō, Harigīt, Jhūḷaṇā, Savaiyā, Roḷā, Kaṭāv, Manhar, and Ghanākṣarī. Pronunciation-dependent, sung, rhyme, and flowing-text details are labeled provisional instead of being quietly guessed.

The faint numbers at the end of each line/pāda show its syllable and mātrā totals. The small counter below the editor shows the totals from the beginning of the current line/pāda to the cursor.

Find a word that fits

Put the cursor on a Kannada or Sanskrit word. When useful choices exist, a small Synonyms bar appears below the slate; otherwise it has the good manners not to occupy the room. Open it to choose a sense and compare each candidate’s syllables, mātrās, and Guru–Laghu pattern. Same rhythm means the whole pattern matches the current word; same mātrās means only the duration total matches.

Selecting a candidate replaces only the word under the cursor and immediately reruns meter analysis. Amarakośa suggestions are dictionary base forms, so their gender is shown and the poet must still supply the right case ending—grammar has declined our offer to become optional. Sanskrit suggestions are rendered in the current Indic script or exact Roman scheme. Kannada suggestions retain Kannada script.

Everything is searched offline. Kannada candidates use Alar’s explicit shared sense identifiers; Sanskrit groups come directly from the Amarakośa. Their separate licenses and provenance are in the data notices.

Teach Chandas your own cadence

If a poem has a recurring shape that is not in the traditional catalog, choose Learn this pattern beside the meter suggestions. Chandas compares every stanza with the active stanza's number of lines, then shows what it actually found: syllable and mātrā ranges, Guru/Laghu endings, possible grouping and yati, repeated refrains, and second-letter or end-rhyme schemes. Stanzas with a different shape are counted as ignored, not stirred into the arithmetic and hoped for the best.

Give the form a name and choose Exact to preserve the observed line patterns, Balanced to enforce the stable parts, or Loose to allow a little movement. Cadence, yati, rhyme, and refrain remain separate checkboxes because a computer should not promote a coincidence to poetic law without asking. The saved result appears in the normal meter selector and checks each stanza like any other selected form.

Custom forms stay on this device and are included in Full backup. They describe the written rhythmic skeleton; they do not yet hear performance timing, stress, melody, or the poet's strategically raised eyebrow. A later, optional performance layer is on the roadmap.

Choose a meter

You may choose a meter before typing anything; it waits patiently and applies to the first stanza when you begin. Chandas.org also suggests likely meters while you type. Select one to check the active stanza, or use Choose any meter to search the full catalog. Ordinary Roman spelling works even when the displayed name has scholarly diacritics.

Suggestions distinguish an exact verse, an exact pāda in an unfinished verse, a strong prefix, and an early possibility that has not supplied enough structural evidence yet. Untyped syllables are progress, not mistakes. The small Common marker only settles close ties; it cannot promote a weaker fit over a stronger one. The best fit remains visible while the rest of the candidates scroll beneath it.

Fixed vṛttas understand the usual four-pāda layout and the compact two-line layout with two pādas on each line. When one line could be either a longer pāda or two shorter ones, Chandas first prefers the reading that invents no boundary; a matching second half-verse can then settle the argument. It will infer four pādas, not repeat a convenient pattern until everyone has gone home.

At a genuine pāda ending where the chosen meter permits it, a final Laghu may fill a signature's final Guru position. Chandas keeps the written syllable Laghu, shows its effective metrical value as Guru, and does not paint it red; the exception is never borrowed by an internal syllable. The meter—not Kannada, Telugu, Gujarati, or Devanagari script—decides the rule. This is why Kālidāsa's Meghadūta opening scans cleanly as Mandākrāntā even though its third and fourth pādas end in short षु.

Anuṣṭubh is equally happy with four eight-syllable lines or the common two-line arrangement with two pādas on each line. To keep early suggestions calm, the two-line reading becomes an exact automatic match only after both sixteen-syllable half-verses are present. Select Anuṣṭubh sooner if you want guidance while writing the second half. Its template still shows four pādas, because the meter has not misplaced two of them merely to save screen space.

Gujarati Caupāī, Doharō, Soraṭhō, and Roḷā accept separate caraṇas as well as the familiar two-line arrangement. Their Ghost templates still show every caraṇa, because compact printing should not make half the meter vanish. Gujarati-only entries are suggested automatically only for Gujarati script, while the full catalog remains searchable by ordinary Roman spelling.

Odd pādas may use pathyā or the standard classical na-, bha-, ma-, and ra-vipulā cadences; even pādas retain their fixed cadence. Ma- and ra-vipulā need a visible word or punctuation break at their caesura. Rare early variants remain outside this conservative rule set until they have a reviewed corpus.

Each blank-line-separated stanza may have its own meter. Use Clear selected meter when you would like the red ink to take a tea break.

Kannada Kanda is listed separately from Āryāgīti. Chandas checks its four lines, sixteen four-mātrā gaṇas, special sixth and fourteenth gaṇas, required Guru endings, conditional yati, and dvitīyākṣara-prāsa. A written final Laghu may fill the required Guru only on lines 2 and 4; lines 1 and 3 receive no such pardon. The rule set remains marked as possible—not exact—until its golden corpus and historical prāsa variants are reviewed.

The quantitative Kannada Ṣaṭpadis Śara, Kusuma, Bhoga, Bhāminī, Parivardhinī, Vārdhaka, and Uddaṇḍa allow the same final-Laghu adjustment only on their longer third and sixth lines. Native Telugu meters keep their own endings instead of inheriting a rule merely because the letters happen to be Telugu.

Pañcamātrā Chaupadi is available in the familiar Mankuthimmana Kagga form. Chandas checks the four-line written frame as 5555 / 5553 / 5555 / 5551, opening dvitīyākṣara-prāsa, and ending antya-prāsa, while keeping pādānta lengthening, śithila-dvitva, and broader historical variants visibly provisional. In other words, it knows this Kagga gait without claiming that every four-legged creature is the same horse. The broader historical family is also available in separate mātrā Chaupadi (5553 / 5553 / 5553 / 5553) and aṃśa Chaupadi (VR on each line) entries, so choosing one never silently changes the other.

Utsāha, Mandānila, and Lalita Ragaḷe are repeatable-line meters, not four-line stanzas wearing an unnecessarily formal hat. Chandas checks every written line, pairwise end-rhyme, and the documented Mandānila alternatives without reporting unwritten future lines as missing. These rules remain provisional pending expert corpus review. Utsāha accepts both its four-gaṇa 12-mātrā line and eight-gaṇa 24-mātrā line.

The six quantitative Ṣaṭpadi forms—Śara, Kusuma, Bhoga, Bhāminī, Parivardhinī, and Vārdhaka—are checked as complete six-line verses. Lines three and six use the longer form; merely arriving with six lines is no longer enough to borrow a Ṣaṭpadi name tag. Uddaṇḍa's 5 / 5 / 8 four-mātrā-gaṇa frame and the older aṃśa Ṣaṭpadi VV / VV / VVR frame are separate choices.

Classical Tripadi, Eḷe, Sāṅgatya, Sobagina Sone, the Akkara forms, and several older forms use aṃśa-gaṇas. Their references use B for Brahma, V for Viṣṇu, and R for Rudra. Chandas checks the documented core frames while keeping melodic substitutions and prāsa visibly provisional.

The compact Scansion menu controls the faint marks drawn over the writing slate. Auto shows realized V/B/R boundaries for a selected aṃśa meter; Mātrā gait 3+5 and 5+3 instead show alternating rhythmic boundaries on every authored line. Each gaṇa or number is centred above its own letter group while its faint vertical separators remain independent. Any leftover cadence appears as x=n beside the line totals. Only one set of boundaries appears at once—because a poem can have layers, but the toolbar need not become an onion farm.

A dashed gait boundary means the target beat falls inside a Guru. Chandas keeps that Unicode syllable whole and treats the mark as an aid to recitation, not a red violation. These faint labels never enter copied, shared, or saved verse text.

Chandovatamsa keeps Nāgavarma's VVVB as its canonical signature. A poet-reviewed realization may use VRVB on line two and VVVV on line three; these contracted substitute gaṇas are accepted without inventing a karṣaṇa mark. The live analysis shows the realized gaṇa scan separately, so the family recipe and tonight's singing do not get mistaken for twins.

The historical choices are deliberately named: Nāgavarma's Chandovatamsa is VVVB, while Jayakīrti's VBBBB form appears as Ādivarāha. Nāgavarma's Akkarike is VBVBVR, and his primary Madanavatī form is VVVVV+G. This saves one ancient disagreement from becoming a modern dropdown argument.

In the standard realization of these classical aṃśa meters, a Laghu in any aṃśa after the first is pulled by one mātrā in recitation. Chandas leaves it Laghu and adds a faint superscript ಽ: for example, Viṣṇu GLL reads as G Lಽ Lಽ. The opening G or LL is not marked. Select the meter for its recital guide; Chandas may also show it when exactly one complete aṃśa meter is detected. If gaṇa division is ambiguous, only marks common to every best reading are shown.

Folk Tripadi (Kannada) is a separate choice with the later mātrā gait 5555 / 5455 / 545. When a gaṇa-final written Laghu needs one extra beat in recitation, a faint superscript ಽ shows the inferred sung mātrā. The mark is a helpful little hitch in the voice, not an uninvited character in your poem: copying and sharing retain only what you typed. Classical Tripadi keeps its stricter aṃśa rules. A third, explicitly historical mātrā Tripadi entry uses 5555 / 5355 / 535; the three forms do not borrow one another's exceptions.

The fixed-vṛtta catalog also includes the Kannada Campakamāle and Mahāsragdharā signatures. Each replaces the initial Guru of its Utpalamālikā or Sragdharā counterpart with two Laghus, as described by the cited Kannada prosody formula.

Saved poems and backups

New now means “put this poem safely on the shelf and give me a clean slate,” not “feed it to the void.” Open Saved poems in the header to search by title or verse, reopen, share, rename, duplicate, or delete work stored on this device. A saved poem's Share button offers the same text, meter, link, social, and analysis-link actions as the writing slate without replacing the poem currently being edited.

This library has no cloud database and no account. Browser or app data can still be cleared. Save all to .txt creates a readable UTF-8 text file containing every title, edit time, selected meter, and poem; send it to Files, Drive, or another destination offered by your device. It is for reading and copying, not exact restoration. Full backup downloads the JSON file that preserves templates and editor state, and Restore full backup imports it. Restore preserves both versions when an ID collision contains different writing. Poetry may clash; backups need not.

Open a verse from a link

Put a URL-encoded verse directly after the question mark: https://chandas.org?<verse>. For links that also choose guidance, use ?verse=<verse>&meter=madhu&template=strong. text is accepted as an alias for verse, and template may be ghost or strong. Meter IDs, displayed names, and ordinary Roman spellings without diacritics are accepted.

If a local draft already exists, the linked verse is appended as a new padya rather than evicting the resident poet. The query is removed after import, so Reload does not develop an enthusiasm for adding the same verse repeatedly. Strong guidance falls back to Ghost when that meter family has not completed Strong-template review.

The Share dialog's Copy analysis link button builds this link for you. It preserves the authored verse and each stanza's selected meter and Ghost/Strong choice using numbered parameters such as meter1 and template1. Empty Strong positions stay model state rather than authored characters, while filled out-of-order positions retain their slots. Leading and interior blank lines are preserved too, so a samasyā-pūraṇa link containing only its final pāda does not promote it to line one. Guide symbols stay out of the link. This is the honest, plain URL version, so a generous poem may produce a URL with similarly generous proportions.

A verse in a URL may be recorded in browser history, messages, or hosting request logs before Chandas opens it. Use link import only for text you are comfortable placing in a link; ordinary editor drafts continue to stay on the device.

Choose Ghost or Strong guidance

After selecting a meter, enable Show template for a faint guide. It shows the whole logical verse even if you have typed only one line: four lines for fixed vṛttas, Anuṣṭubh, and Kanda; six lines for a Ṣaṭpadi; or the shape declared by another meter. Existing lines also show their remaining inline hint.

Fixed meters show Laghu/Guru positions, Anuṣṭubh shows fixed and variable positions, mātrā meters show group capacity and progress, and aṃśa meters show their B/V/R frame. A two-pattern vṛtta alternates its patterns across the four lines rather than quietly forgetting lines three and four.

The guide is only a guide: it never becomes part of your poem, analysis, copy, or shared text. Ghosts, in this case, are unusually respectful of boundaries.

For fixed vṛttas, choose Strong to get one fillable box per Guru/Laghu position. You may begin anywhere—even line four, position seven, if inspiration has taken the scenic route. A red box disagrees with the expected weight; an empty box remains only interface state and is never inserted into the poem.

Strong boxes follow the visible syllable onset, so a conjunct stays shaped as ಪಾ · ರ್ಥಾ or पा · र्था, not ಪಾರ್ · ಥಾ or पार् · था. The metrical scan still crosses the box boundary: a short syllable before a following conjunct is correctly Guru even though the conjunct is displayed intact in the next box.

Strong drafts, including deliberately empty positions, recover on this device. Switching back to Ghost keeps the authored text and meaningful blank rows. Copy and Share include only what you typed, plus the line breaks needed to keep a later pāda later; analysis links also retain every empty Strong slot. Structural, mātrā, and aṃśa meters keep Ghost guidance until their strong layouts have completed the same rule and corpus review as their detectors.

Optional śithila-dvitva detection

Old Kannada poetry may pronounce some written conjuncts lightly: a preceding short syllable that would ordinarily become Guru can remain Laghu. Enable Detect śithila-dvitva in the analysis panel when you want Chandas to test that reading. It is deliberately off for every new poem; the ordinary analyzer is not feeling adventurous without permission.

Historical ೞ is a real Kannada consonant despite its mistaken original Unicode name “KANNADA LETTER FA” (its formal alias is KANNADA LETTER LLLA). Chandas recognizes it alongside ಲ, ಳ, and ಱ even when this option is off. Thus ಕೞ್ದ ordinarily scans Guru–Laghu; the optional pass may realize it as Laghu–Laghu when the meter supports śithila-dvitva.

The first conservative pass recognizes reviewed repha and lateral environments involving ರ್, ಱ್, ಳ್, and historical ೞ್ before eligible simple consonants. It adopts a Laghu realization only when that improves the metrical fit, recomputes the complete analysis, and marks the used conjunct with a faint *. Similar-looking conjuncts are not automatically equivalent; lexical and historical exceptions remain a matter for corpus review. The initial rule is informed by Venkatachala Shastry's Old Kannada grammar discussion; it remains deliberately conservative rather than pretending every repha has received a tiny poetic pardon.

The choice is remembered with that saved poem. An analysis link includes sd=1 only when detection is enabled. Neither the asterisk nor the alternate analysis changes the authored poem, copy output, or readable text export.

Read the prāsa colors

Whenever a stanza is written in Kannada script, Chandas automatically checks the terminal consonant of its second akṣara across the available lines—even when no meter is selected or the selected meter comes from the Sanskrit vṛtta catalog. This automatic check is advisory: it helps with an ad-hoc meter without declaring the selected meter wrong.

A cataloged Kannada meter may make the same prāsa a required part of its validation and may add special positions of its own. In either case, the first syllable’s Guru/Laghu weight must agree. A teal letter highlight means the prāsa matches, red means the consonant differs, and amber means the opening weight differs. The usual dotted Laghu or solid Guru underline remains on the highlighted letter.

Ottu letters are compared by sound but colored as one safely shaped cluster. So ಮಲ್ಪ and ಜಂಪ meet on ಪ, while the editor may color all of ಲ್ಪ. Spaces and punctuation politely stay out of the argument. Matching final consonants are checked for meters such as Ragaḷe and Chaupadi that declare antya-prāsa.

Each nonblank line is treated as a pāda for the automatic check; danda and pipe marks can divide more than one pāda on a line, and a blank line starts a fresh stanza. At least two eligible Kannada pādas are needed. Mixed Kannada–Devanagari stanzas are not compared across scripts. Danda, double danda, verse numbers, Markdown marks, and other punctuation are script-neutral.

Chandas also reports Ādi-prāsa found when the first consonant of every available line matches. That is a pleasant finding, not a scolding: its absence is not an error unless a future meter rule explicitly requires it.

Save, copy, and share

Your current draft and per-stanza choices are recovered privately on this device. Copy returns the poem without markings. Share opens the device or browser share options; Chandas never posts without you.

The core editor works offline after installation or a successful web visit. Meter suggestions remain guidance, especially for traditions and optional rules not yet included in the catalog.