Eleven Sanskrit texts, each verse tied to the second it is chanted in a recorded upanyāsam. Click a verse, hear it. Eight stand on editions of their own; the other three are kāṇḍas of one nibandha — a digest made of quotations, indexed by the authority it credits as well as by the text. Two further texts have no edition to hang them on, and one month of Māgha talks is indexed by the thirteen books it draws on.
They are built to argue for two things the Purāṇas still lack: recitation that is addressable at the verse, and an index of names across the corpus. Read the two proposals →
Not an atlas · a text with no edition
Tulā Kāverī Māhātmya · Āgneya Purāṇa
Two reciters, recorded years apart, chant 141 of the same verses in the same order — and those verses appear in none of the 488,149 we index. No printed or digital exemplar has been found, so there is nothing here to align a recording to. The page therefore does what is left, and it turns out to be the more interesting thing: it sets one recording against the other, prints both decodes where they differ, and lets you hear the same verse in two mouths.
Not an atlas · an index to a recording
Śrāvaṇa Māsa Māhātmya · one reciter, a month of days
A vrata a day for a whole month, on a text that is in none of the 488,149 verses we index and has no second reciter to set against it. So there is neither an edition to align to nor a witness to compare with, and what is left is an anukramaṇikā: 959 passages of chant, each placed in its recording to the second by forced alignment, so a month of Śrāvaṇa can be heard without being read. On 15 of the days the reciter steps into a text we do hold — a stotra entire, a consecutive run of the Kāśīkhaṇḍa — and those are marked and cited.
Nobody typed these timestamps. Each recording was transcribed, the transcript matched against the printed text, and every matched verse then force-aligned to the waveform — so an onset is a measurement of the audio, not an editorial guess. Where a verse is not marked, that is a claim too: the speaker did not chant it, or the pass could not find it. The atlas distinguishes those cases rather than smoothing them into one.
A lane is one traversal of the text, and what separates two lanes is not the same thing in every atlas. In Kārtika they are reciters: Speaker 1 reads the book straight through, while Speaker 2 follows the calendar, taking each day's verses from wherever the text supports them. Together they reach more of the work than either alone, and the verses both chant are the same words in two voices.
In the Adhyātma Rāmāyaṇa the two lanes are one reciter in two settings. Upanyāsa is seven days of discourse, a few verses at a time with their explanation between them. Pārāyaṇa is seven days given to near-continuous recitation before any exposition begins — which is why it reaches 3,974 verses where the same reciter's seven days of discourse reach 722. Reciting and expounding are different undertakings, and the two lanes are the clearest measurement of that on this site. Verses are heard in both: the same words, in the same voice, twice.
Playback is a YouTube embed, so the page needs the network and a browser
that will run youtube.com/iframe_api. Everything else — the
text, the divisions, the measured timings — is served from this site and
works offline once loaded.
The Sanskrit is set in Noto Serif Devanagari, which the site serves itself — you do not need a Devanagari font installed, and the page asks no third party for one.
These atlases are the working proof behind two proposals. They share a corpus, and each is worth more with the other: one would make the Purāṇas audible at the verse, the other would make them searchable by name.
Proposal · recording
Commission recitation that is addressable at the verse from the moment it is recorded. Retrofitting reaches 61–88% of a text because a reciter does not chant every verse; that gap is what commissioning buys.
Proposal · index of names
A scholar-verified index of the names, peoples and places of the Paurāṇic corpus — what Sorensen did for the Mahābhārata, beginning with the Harivaṃśa, the largest unindexed book in Sanskrit literature.
Working note · not a proposal
Both proposals ask for scholarly reading time. This note describes the instrument that reading time is spent inside — why a published page is derived rather than edited, and why an index of this kind cannot be a wiki. It turns on two passages in the Kārtika atlas: Satyabhāmā asking Kṛṣṇa what she did in a former birth, and a brahmin promised he will be reborn as Daśaratha. Both stories are retold widely and addressed almost nowhere — and the corpus names both in its own chapter headings.