27 October: a spotlight on AI Jam Sessions, interlock and XRPL Camp

Today’s Colab work produced little to report, so this issue is a spotlight: three of the studio’s tools that are public and shipping, what each one does, and what changed in it lately.

In this issue:

1. AI Jam Sessions: an AI that plays, sings and listens
2. interlock 1.0.8 to 1.0.10
3. XRPL Camp: the XRP Ledger in one sitting
4. A short note from today’s runs

AI Jam Sessions: an AI that plays, sings and listens

AI Jam Sessions is an MCP server that teaches a language model to play piano and guitar, and to sing. A model can read text, but it has no ears and no fingers, so the server gives it senses it can use:

• Reading: 109 songs across 12 genres, every one annotated with its history, its structure bar by bar, and what it teaches.
• Hearing: six sound engines, from a sampled piano to a modelled guitar and a vocal tract, playing through real speakers.
• Seeing: a piano roll the model can read back, a guitar tablature editor, and a browser cockpit.
• Remembering: a practice journal that carries over from one session to the next.
• Listening: point it at a recording and it measures what is there, with the same kind of tools it uses on the score.

It has 56 MCP tools in all, and installs from npm as @mcptoolshop/ai-jam-sessions; the current release is 2.8.0.

Three hymns, sung whole

Version 2.8.0 added whole songs sung by a synthetic voice on the piano’s clock: Amazing Grace and America the Beautiful, every verse. The Battle Hymn of the Republic followed on the landing page. The performances were rendered offline on a rented GPU, not by the server’s own sound engines. Each is built from sixteen takes of a singer conditioned on the score, with one take chosen for each phrase and checked for timing and pitch by two instruments each, then passed by ear.

	Amazing Grace	America the Beautiful	Battle Hymn
syllables within 40 ms of the score	107 of 112	217 of 224	285 of 414
notes within 50 cents	135 of 140	218 of 224	368 of 424

The Battle Hymn’s quick dotted notes are left on the singer’s own timing, which is where most of its timing misses come from. The voice is SoulX-Singer (Apache-2.0) with its own example timbre, imitating no real singer, and all three hymns are in the public domain.

Data others can use

Two datasets from the same work are public on Hugging Face:

• jam-actions-v0: 57 multi-turn MCP tool-use traces over four classical piano works, for teaching a model to answer questions about music by calling tools rather than guessing.
• jam-actions-acoustic-v0: 72 records over audio rather than notation, each label checked against what the analysis tools actually return. It ships no audio: each record carries a recipe and the hash of the waveform it produces.

Both carry their licences and sources in their cards. Each was corrected on 25 September when an audit found files whose arrangement licence couldn’t be established, and the records built from them were withdrawn.

interlock 1.0.8 to 1.0.10

interlock picks a model, an engine and a device for a job from speeds it has measured. Issue 31 told how its first runs on a rented T4 card exposed defects in how it measured, fixed in 1.0.5 to 1.0.7. The next three releases came from the same runs:

• 1.0.8: nothing waits for ever. Every cell has a wall-time cap, and a call that reads the card’s memory now has a deadline: one run had gone silent for 56 minutes waiting on it. The harness prints a line at each step, so the next hang names its own call. Rig checks get their own identity instead of borrowing a Colab card’s name. And interlock route --concurrency N ranks results for the number of requests you’ll actually send.
• 1.0.9: a hung cell is cleaned up. When a cell’s deadline fires, the server is restarted and any late answer from the abandoned cell is refused. The fallback reading of the card’s memory is checked strictly, and an answer it can’t trust is recorded as unmeasured.
• 1.0.10: noise is recorded, not used to discard. A check meant to catch a cached prompt was also discarding real cells, because the time to read the prompt varies from pass to pass. Each measured row now records that spread instead, and the cells are kept.

interlock 1.0.10 is on PyPI as mcptoolshop-interlock and needs Python 3.13 or newer.

XRPL Camp: the XRP Ledger in one sitting

XRPL Camp teaches the XRP Ledger by doing it: real transactions on the ledger’s test network, in about ten minutes, ending with a proof of what you did that anyone can check against the ledger. It’s on PyPI as xrpl-camp and npm as @mcptoolshop/xrpl-camp, both at 1.5.2.

• 1.5.0 wired in the last parts of the course: a walkthrough of the reserve an account owes for the things it owns, a plain-text certificate you can keep and share, and dated receipts each time a proof is verified.
• 1.5.1 and 1.5.2 tightened the edges. A submission whose result comes back missing is no longer taken as a success: the camp looks the transaction up on the ledger and reports what it finds. The reserve step says plainly when a change couldn’t be measured yet. And the standalone downloads no longer bundle a leftover install record from their build.

A short note from today’s runs

On today’s T4 runs, the Colab machine’s memory limit stopped some of the largest cells: when a model’s working cache spilled out of the card into the machine’s memory, the limit killed the model server at about 11.5 GB of its own memory. In three runs the process running our harness was killed a moment later as well (two seen directly, one inferred from its logs); the rows written up to that point were kept, but the rest of that step was lost. The run launcher now notices a dead run and stops it within about two minutes, instead of waiting out its full time. A check that refuses such cells before they load is being built for interlock 1.0.11.