This site is an ongoing labor of love — digitizing, cleaning up, and sharing the live recordings of Jerry Hannan, the Mad Hannans, and Sean Hannan. Here’s how it has come together so far.
Week one — getting it online (June 9–13, 2026)
The archive went live as a simple homepage and quickly grew a proper foundation.
- Put up the first “About This Archive” page explaining what these recordings are and why I’m sharing them.
- Password-protected the lossless WAV masters so the big files stay controlled, while keeping the music easy to listen to.
- Set up automatic publishing — every change now deploys itself to the web.
- Rebuilt the site around a data-driven generator, giving each show its own page with a proper track list.
- Posted the first track-listed shows: Jerry Hannan at 19 Broadway (Jan 8, 2001), the Mad Hannans at Sweetwater (Feb 17, 2000), and Sean Hannan at 19 Broadway (Jan 24, 2000) — followed by a redesign of the archive and per-song sharing links.
Week two — more shows, more polish (June 14–20, 2026)
The site split into proper sections — Home, Archive, Shows, Updates, and Contact — and an Updates feed began tracking changes as they happen. New shows kept arriving:
- Sean Hannan at 19 Broadway (Feb 21, 2000)
- Jerry Hannan at 19 Broadway (Jan 15, 2001) — later volume-normalized across all 31 tracks
- The Mad Hannans at Sweetwater (Oct 17, 2000) — the Birthday Show, with a full lossless download
- Jerry Hannan at 19 Broadway (June 21, 1999)
I also added notes about audio levels and the purpose of the archive, and began marking which recordings came straight off the soundboard (SBD) versus from the audience (AUD).
Week three — listening, searching, restoring (June 21–26, 2026)
The biggest leap in how the archive feels to use:
- Every full show now streams as a clean 320 kbps MP3 straight from the show pages.
- Added the Sean Hannan 19 Broadway (unknown date) set and the Mad Hannans at Cafe Java (Sept 9, 1999).
- Introduced waveform players so you can see each song as you listen.
- Added an Archive sort toggle (browse by date or by artist) and a search engine to find any song or show by title, artist, venue, date, or tag — backed by an ongoing effort to tag covers and songwriters.
- Posted the Mad Hannans at Sweetwater (Jan 6, 2001) — a rough, muffled audience tape that I then went back and cleaned up with EQ and loudness normalization to make it easier on the ears.
- Gave the Sean Hannan (Feb 21, 2000) set the same loudness treatment, and corrected a long-mislabeled song: “Irish Song” is really “Ode to Biddy McGee.”
Week four — a real audio pipeline (June 27–28, 2026)
The cleanup work grew up into a proper, repeatable process. Instead of treating each show by hand, there’s now an automated audio-engineering workflow that takes the original lossless masters and brings every track to one consistent, comfortable listening level across the whole archive — solo sets and full-band shows alike — all with a ceiling that prevents digital clipping.
- Each processed show now carries a “Technical data” panel — the before-and-after loudness, true peak, and dynamic range for every song, plus exactly which version of the workflow touched each track.
- Every show page shows its processing status at a glance — whether it’s been through the new workflow yet, or is still on the list.
- Re-mastered to the new standard: Jerry Hannan at 19 Broadway — June 21, 1999; Jan 8, 2001 (now with lossless FLAC downloads added); and Jan 15, 2001 (redone from the masters to the new loudness standard).
- Settled on one loudness standard for the entire archive — after A/B testing confirmed the full-band shows gained nothing audible from being louder, the Mad Hannans sets now sit at the same comfortable level as everything else. The earlier −16 passes are being redone to match. First through the new process: the Mad Hannans at Sweetwater (Feb 17, 2000), the Oct 17, 2000 Birthday Show, the Cafe Java set (Sept 9, 1999), and Sean Hannan’s Feb 21, 2000 set.
- Added corrective EQ to the workflow for salvaging poor source tapes, and used it to rebuild the muddy Mad Hannans at Sweetwater (Jan 6, 2001) straight from the raw transfer — cutting the low-mid mud and lifting presence and air to bring the vocals forward, then normalizing to the −20 standard. The exact EQ is recorded with each track.
- Behind the scenes: faster page loads, better link previews when a show is shared, and more reliable, properly protected lossless downloads.
Week five — repairing the damaged peaks (June 29–30, 2026)
Some older recordings have a flaw that volume adjustment alone can’t fix: bursts of applause so loud they overloaded the original tape and distorted. This week the focus turned to repairing that damage by hand before normalizing.
- Went back to one of the very first shows posted — Sean Hannan at 19 Broadway (Jan 24, 2000), a 31-song solo set — and found nine tracks where loud audience clapping had clipped into distortion. Each one was hand-edited in Audacity to smooth out the overloaded peaks, then the whole show was brought to the archive’s −20 loudness standard.
- Brought the undated Sean Hannan at 19 Broadway set (18 songs, with guest turns from Jerry Hannan and Kelly Peterson) through the same workflow — replacing an earlier −16 pass with a clean −20 normalization from the source.
- Kept the whole archive in sync: the original masters, the hand-edited versions, and the final normalized tracks are now all preserved together in the cloud backup.
Week six — New George’s, more shows, and a technical clean-up (July 5–7, 2026)
Added the Mad Hannans at New George’s in San Rafael (Oct 13, 1999) as 14 individual tracks. The show survives as two audience DAT transfers, so I A/B’d them side by side and went with the one that sounded best. Each song was split out, the applause gently faded at the end for easier back-to-back listening, and the whole set brought to the archive’s −20 loudness standard. The set opens with a short soundcheck.
Also launched a new “Songs” section — a cross-referenced matrix of every song against every show. Because the same songs turn up again and again across Jerry, the Mad Hannans, and Sean, you can now pick a song and see every time it was played, each with a player and a link straight to that performance. Browse it as a filterable list or a songs-by-shows grid, sort by how often a song was played, filter by artist, and search for any song by name.
Added two more early Jerry solo shows at 19 Broadway in Fairfax — May 10, 1999 (21 songs) and July 19, 1999 (a long, 26-song set). Each was split into individual tracks by hand in Audacity, with the applause studio-faded at the end of every song and a little clipping cleaned up on the loudest peaks, then run through the pipeline to the archive’s −20 loudness standard. All 47 new tracks flow straight into the Songs matrix, which now also treats Jerry’s “My Dear” and “Ride On” as the same song.
Then a stretch of polish rather than new tapes. The Songs page got a real workover: the list is tighter and easier to scan, with each artist shown as a small colour-coded dot lined up in fixed columns — so you can run your eye down and see at a glance which songs Jerry, the Mad Hannans, or Sean played. The songs-by-shows grid now scrolls sideways properly, and its artist filter finally hides the other artists’ columns instead of just their rows. Two songs on the Sweetwater soundboard (Feb 17, 2000) were corrected as well: a track listed as “There She Was” is really “Butter,” and the copy here was missing its opening chords, so it was swapped for the complete take; and a stray “Da Da Da” was reunited with its full name, the medley “The Kiss – Da Da Da.” Each fix was pushed everywhere the audio lives at once — the streaming store, the cloud archive, and the working files.
Under the hood, the site grew up a little. Every page now carries structured data so search engines read these as live music recordings, performances, and songs rather than anonymous pages. There’s a proper dark mode that follows your device’s setting, an RSS feed of these updates so you can follow new additions, and a round of accessibility work — visible keyboard focus, a skip-to-content link, reduced-motion support, and stronger text contrast. Plumbing you never see, but the kind that keeps an archive healthy.
Week seven — a playlist generator (July 7–9, 2026)
A new way to listen to the archive: build your own set instead of working through one show at a time.
- Every curated track now carries a proper tag vocabulary — mood, style, and format — plus a dedicated songwriter credit, separating Hannan originals from covers (named by writer) and Irish traditionals.
- Launched Playlist: filter the whole archive by artist, venue, audience/soundboard, and any combination of tags, choose how long you want to listen — a number of songs, a number of minutes, or endless shuffle — and hit play. However you slice it, the queue never plays the same song twice, even one with a dozen different live takes across the years.
- Every playlist is shareable. The exact set of songs lives in the link itself, and a “Copy share link” button hands you a short /play/… URL to send along.
Also added the Mad Hannans at 4th Street Tavern in San Rafael (May 1, 1999) as 24 individual tracks — split, faded, and clip-repaired by hand before the usual pass to the archive’s −20 loudness standard. The set opens with a short soundcheck.
Added a Sean & Jerry Hannan show at 19 Broadway from December 1999 (exact date unknown) as 30 individual tracks, split from the audience DAT and run through the usual pass to the archive’s −20 loudness standard. The tape itself turned out to be recorded at 32 kHz throughout — despite one of its two source files being labelled “48KHz” — so a handful of tracks that had been exported at the wrong rate were corrected to match before processing.
Added a marathon Jerry Hannan solo show at 19 Broadway (March 29, 1999) as 34 individual tracks — the biggest single-show track count on the site so far, mixing Hannan originals with a run of John Prine covers, two Cat Stevens songs, and a closing stretch of Irish traditionals. It’s also a first for the archive’s audio work: broadband noise reduction was applied across the whole tape in Audacity before the songs were split out, taking the hiss floor down dramatically, and that step is now recorded in each show’s technical data as a “pre-edit” alongside the usual fades and clip repairs.
Added a Sean Hannan solo show at 19 Broadway from November 1999 as 21 individual tracks, split from the audience DAT and run through the usual pass to the archive’s −20 loudness standard. The set pairs Sean’s originals with a run of covers — Bob Dylan, Tom Petty, Cat Stevens, John Prine, and the Beatles — alongside a couple of Irish traditionals.
The whole journey a tape makes — from DAT transfer through hand-editing to the published show page — is now documented step by step on The Process.
Week eight — a new coat of paint (July 10–11, 2026)
Mostly a design week: the site keeps the same bones, but reads and feels more like one place now instead of a homepage bolted onto everything else.
- Gave the whole site a type and color refresh — Cormorant Garamond for headings, Karla for body text, and a touch of monospace on labels and data, all built on the archive’s own existing warm paper and forest-green palette rather than a new one. The Process keeps its own plain, reading-focused layout, the same as The Manual.
- Rebuilt the homepage around a real grid of recently added shows instead of a wall of text up front. “Play a random tape” genuinely picks one and jumps to it; the “Why I’m doing this” and “What’s here” sections keep every word they always had, just tucked behind a “Read more” so first-time visitors aren’t met with an essay.
- A pass on accessibility: every play button now announces the actual song, artist, and date to a screen reader instead of a bare “Play” repeated on every row, and the seek bar is a real, keyboard-operable control rather than a mouse-only strip.
- The Songs index got much lighter — performances used to all load at once (well over 400 of them); now a song’s recordings load in only when you actually open it.
- The Archive page explains its own shorthand now — a small legend for AUD, SBD, NR, and the track-count mark.
- The Playlist builder gained three one-click starting points (a 45-minute mixed set, Traditional & Irish, and Soundboard recordings), a clear-filters button, and an estimated length before you commit to generating one.
Added a Jerry Hannan solo show at 19 Broadway from February 1, 1999 as 21 individual tracks, split from an audience tape with some significant dropouts — a couple of tracks (Candle in the Wind, Angel of Montgomery) show the effect — and Sean Hannan sitting in for a few songs. Perfect Autumn Day turns up twice in the set, still a work in progress this early on. The whole tape got a pass of broadband noise reduction (6dB, sensitivity 5) in Audacity before the tracks were split, then the usual −20 LUFS normalization.
Added a Mad Hannans show at the Sweetwater in Mill Valley from May 18, 1999 as 22 individual tracks — almost all originals, plus covers of Gary Lee and the Showdown’s “The Rodeo Song,” Lefty Frizzell’s “Long Black Veil,” and a singalong “Rocky Road to Dublin.” This one's also a first: rather than noise reduction, the whole tape got an EQ pass in Audacity to brighten up a muddy room sound, so the archive’s pre-edit badge now distinguishes “noise-reduced” from a plain “pre-edited” tag for tapes treated some other way.
Week nine — the Cafe Java run, a subtle audio bug, and hand-picked playlists (July 12–15, 2026)
A run of Jerry’s 1999 Cafe Java residency in Larkspur arrived — and processing it surfaced the archive’s subtlest audio bug yet.
Added a Jerry Hannan solo show at Cafe Java in Larkspur from April 29, 1999 as 21 individual tracks — mostly originals, with covers of John Prine’s “Illegal Smile,” Cat Stevens’ “Father and Son,” Paddy Nash’s “Model Family Man,” and Mentor Williams’ “Drift Away,” plus a pair of Irish traditionals. The set closes out with “The Blarney Stone Blues.” No noise reduction or EQ needed on this tape — just the usual hand-edits and −20 LUFS normalization.
Added another Jerry Hannan solo show at Cafe Java in Larkspur, this one from March 25, 1999, as 21 individual tracks — originals plus covers of John Prine (including a rare “Angel of Montgomery”) and Paddy Nash, a pair of Irish traditionals, and a guest turn from Kelly Peterson on “Daddy.” The tape also holds a second date (April 8, 1999) split out as its own show. This one got a pass of broadband noise reduction (8dB, sensitivity 5) in Audacity before the tracks were split, then the usual −20 LUFS normalization.
Added the other half of that tape: a Jerry Hannan solo show at Cafe Java from April 8, 1999, as 23 individual tracks — originals alongside covers of Cat Stevens, Bruce Springsteen, Lefty Frizzell, John Prine, and Van Morrison, closing with the Irish traditional “The Parting Glass.” Kelly Peterson again sits in on “Daddy,” and the Eagles’ “Peaceful Easy Feeling” makes its first appearance in the archive. Straightforward hand-edits and −20 LUFS normalization, no NR or EQ needed.
Added one more Jerry Hannan solo show at Cafe Java in Larkspur, this one from May 27, 1999, as 20 individual tracks — mostly originals, with covers of John Prine’s “Illegal Smile” and “Please Don’t Bury Me,” and Paddy Nash’s “Model Family Man,” closing with the traditional “The Parting Glass.” Processed straight through the corrected pipeline from the start, no reprocess needed.
Added the biggest set of the residency yet: a Jerry Hannan solo show at Cafe Java in Larkspur from June 17, 1999, as 27 individual tracks — mostly originals, with covers of John Prine (“Illegal Smile,” “Angel from Montgomery”), Cat Stevens’ “Father and Son,” Paddy Nash’s “Model Family Man,” the Eagles’ “Peaceful Easy Feeling,” Bruce Springsteen’s “Tougher Than the Rest,” and Peter Jones’ “Kilkelly Ireland,” plus a pair of Irish traditionals. This tape ran hot: the first diagnose pass came back with more than half the tracks flagged for audible clipping, so the fades were touched up and the clipping tamed by hand in Audacity before a clean second pass and the usual −20 LUFS normalization.
Added a Jerry Hannan solo show at 19 Broadway in Fairfax from November 15, 1999, as 14 individual tracks — a shorter set of mostly originals, including an early version of “Burning Man” (long before the song found its familiar shape) and a short alternate-verse take of “Hollywood” right after the full version, alongside covers of Bruce Springsteen and Steve Poltz. Split, faded, and clip-repaired by hand, then normalized to the archive’s −20 LUFS standard.
Added a much longer Jerry Hannan solo show at 19 Broadway from October 25, 1999, as 26 individual tracks — mostly originals, including an alternate take of “ABC” and an early version of “Hollywood,” alongside covers of Bruce Springsteen, Steve Poltz, John Prine, the Eagles, Paddy Nash, Marty Robbins, and Merle Haggard, plus a couple of Irish traditionals. Kelly Peterson joins as a guest vocalist on two of the covers. The whole tape got a pass of broadband noise reduction (6dB, sensitivity 5) in Audacity before the tracks were split, then the usual −20 LUFS normalization.
Added a Jerry Hannan solo show at 19 Broadway from June 7, 1999, as 29 individual tracks — mostly originals, with repeat outings for “Luxury of Murder” and “Everything Reminds Me of You,” alongside covers of John Prine, Steve Poltz, Peter Jones’ “Kilkelly Ireland,” Lefty Frizzell’s “Long Black Veil,” and Merle Haggard’s “Okie from Muskogee,” plus three Irish traditionals — including “A Bunch of Thyme” — and a birthday singalong for an audience member named Marco. The tape got a pass of broadband noise reduction in Audacity before the tracks were split (10dB on the opening three tracks, 6dB the rest of the way, sensitivity 5 throughout), then the usual −20 LUFS normalization.
Added a Jerry Hannan solo show at 19 Broadway from August 23, 1999, as 14 individual tracks. The two source reels for this show are mislabeled, so this main set comes from the cleaner of the two. Mostly originals, including an early version of “Rugburns” before it found its familiar shape, alongside covers of Steve Poltz’s “Single Life,” Bruce Springsteen’s “Highway Patrolman,” Lefty Frizzell’s “Long Black Veil,” and Paddy Nash’s “Model Family Man.” Straightforward hand-edits and the usual −20 LUFS normalization, no NR or EQ needed.
Recovered 7 more songs from the rougher of those two August 23 reels and added them to the same show page as a separate “Part 1 (Distorted)” batch below the main set — rough listening in places, but archived rather than lost. It includes an alternate take of “ABC,” another early version of “Rugburns,” and three songs new to the archive: “Troy,” “The Kiss,” and “Da Da Da (Slave to an Angel).” Show pages can now carry more than one batch of tracks from different source tapes, each labeled separately in the track list.
Added the earliest Mad Hannans show in the archive: Marin Brewing Co. in Larkspur, from April 1, 1998, as 17 individual tracks — Troy on bass, and “Da Da Da (Slave to an Angel)” still without the “The Kiss” intro it later grew. It’s a rough audience tape, left without EQ or noise reduction so the room sound stays honest to the source — just the usual hand-edits and −20 LUFS normalization. The set includes an early outing for Steve Miller’s “Space Cowboy,” a cover that stayed a Mad Hannans staple for years afterward.
A year later to the day, the band came back to the same room: added the Mad Hannans at Marin Brewing Co. from April 1, 1999 — the venue’s 10th anniversary — as 15 individual tracks. This time “Da Da Da (Slave to an Angel)” opens with its full “The Kiss” intro, missing from the 1998 show. Two audience transfers of the tape survive; the tracks were split from the cleaner one, with the other kept as an alternate full-show download. Usual hand-edits and −20 LUFS normalization.
- Caught and fixed a silent audio bug — then extended the fix. The April 8 show’s hand-drawn fade-outs came out flattened, and the trail led somewhere unexpected: whenever a track couldn’t reach the archive’s loudness target without clipping, the normalizer was quietly switching from a straight volume change to a compressor-like mode that rides the level back up through quiet passages — erasing fades and squeezing the natural dynamics, with no warning anywhere. On these acoustic tapes the dynamics are the performance, so the workflow made truly linear normalization permanent policy: a track that can’t safely reach −20 lands at its own slightly quieter level instead — inaudible as a difference, where a flattened fade is not. That still left one wrinkle: on tapes where the mic sat close to the crowd, a single loud clap could force a whole track quieter than the music itself ever needed. The workflow learned to tell a clap from music by behavior, not volume — a burst that towers over everything around it, right at a track’s start or end where applause between songs always falls, gets its own gentle cap, while the performance underneath keeps its full, untouched gain. The March 25 show became the first already-published show actually reprocessed under the fix, with its audio pipeline’s own paper trail — visible per track on every show page — now showing exactly which treatment each song received and why; the rest of the archive follows from here.
- Three more shows caught up to the fix. The 19 Broadway solo set from Feb 1, 1999, the Cafe Java show from Apr 29, and the Mad Hannans’ Sweetwater set from May 18 had all gone up before the dynamic-fallback bug was found — all three were re-run through the corrected pipeline, 64 tracks in total, each now carrying the same per-track treatment audit trail.
- Removed a duplicate show. A tape filed as “Cafe Java, May 10, 1999” turned out to be a second digitization of the 19 Broadway May 10, 1999 show already on the site — the same 21 songs in the same order, down to the second. The archive is one show smaller and one truth larger.
- Hand-picked playlists. Until now the Playlist builder could only answer “surprise me” — filters and shuffle. Now every track row across the site carries a small “+” button: collect songs as you browse (the selection follows you from page to page), then build a playlist from exactly what you picked. Queues can be pruned song by song, and any playlist can be saved under a name in your browser — the groundwork is in place for the archive’s own curated playlists to appear on the Playlist page.
Week ten — a more trustworthy audio engine (July 16–17, 2026)
Less new tape this week, more making sure the engine behind every track can be trusted down to the last decimal.
- Closed the last gap in linear normalization. Since the fix a few weeks back, a track that couldn’t safely hit the loudness target landed at its own quieter level instead of risking flattened dynamics — but that safety net still asked the normalizer itself to apply the final gain and trusted it to behave. It now computes the exact gain once and applies it as a single, plain volume change, so there’s no longer any step where a silent fallback could sneak back in, even in principle. The dynamics-preserved check that used to run only on the trickiest tracks now runs on every single one, and two new measurements are recorded alongside the usual loudness numbers — catching the case where two tracks share the same average level but one has a much hotter chorus that makes it sound louder.
- Sean Hannan at 19 Broadway, Jan 24, 2000 — the 31-song set whose clipped applause got hand-repaired back in Week five — was re-run through the current engine. While preparing it, one song, “Did You Ever See,” turned out to be missing from the working files entirely, with the two songs after it shifted up a number to cover the gap; it’s been recovered and the numbering set right.
- Sean Hannan at 19 Broadway, Feb 21, 2000 was also re-run through the current engine, and track 5’s title corrected to its real name, “Ode to Biddy McGee.”
- Sean Hannan at 19 Broadway, Nov 29, 1999 turned out to be another show caught by the flattened-dynamics bug and was re-run through the current engine. Checking it over turned up a second, narrower bug: on a track where audience applause (not the music) set the loudness ceiling, the safety check meant to catch an over-loud result was adjusting a number that didn’t actually move the peak, so an over-loud track could slip through uncaught. That’s fixed now too. Track 7, “The Rodeo Song,” has some clipping baked into the original tape that no amount of re-processing can undo — it’s flagged on the site rather than hidden.
Week eleven — tightening up the site (July 18, 2026)
A round of housekeeping on the site itself rather than the tapes: every interior page now reads as one place with the homepage instead of a collection of separately-styled pages, and a handful of small copy and UI details caught up to where the archive actually stands.
- Interior pages match the homepage. The Archive, Songs, Search, Updates, and every show page now share the homepage’s wider layout, readable type, and rounded card look, instead of a narrower, denser style left over from before the homepage redesign.
- Every listed show is now fully indexed. With the one remaining un-split tape held back for later, the “split shows only” filter on the Archive page — and the copy explaining it — is gone, since there’s nothing left to filter. The Songs page and homepage lost a couple of similar notes that no longer applied.
- A closer look at a song’s performances. Hovering a performance on a song page now shows a quick info card — venue, date, recording type, duration, and file size — without having to open the show itself.
- A handful of smaller fixes: a leftover sentence on the homepage that no longer matched the list below it, and a trimmed-down legend on the Archive page.
- Jerry Hannan at 19 Broadway, June 21, 1999 was re-run through the current engine — the original v1 processing had silently flattened several tracks to exactly −20 LUFS at exactly −1 dBTP, the same dynamic-fallback bug caught elsewhere in the archive; some songs land as quiet as −24.7 LUFS to stay truly linear. The fresh hand-edit pass also used slightly different track names than the archive’s established titles, which have been restored rather than overwritten.
Week twelve — another pass through the engine (July 22–23, 2026)
Single-show tune-ups rather than a broad push.
- Sean Hannan at 19 Broadway (undated show) was re-run through the current engine, replacing an older −16 LUFS Audacity pass. The fresh hand-edit export used slightly different track names than the archive’s established titles for two songs — “Don’t Think Twice It’s All Right” and “Angel from Montgomery” — which have been restored rather than overwritten.
- Mad Hannans at Sweetwater, Jerry’s birthday show, Oct 17, 2000 was re-run through the current engine, replacing the original v1 processing. A couple of title fixes came along with it: “Model Family Man” and “Plastic Lemons” had picked up stray trailing spaces, and “The Kiss / Da Da Da (Slave to an Angel)” had drifted to a different separator than the rest of the archive uses — all restored to the established titles.
- Mad Hannans at Sweetwater, Feb 17, 2000 was re-run through the current engine, replacing the original v2 processing. This one took a couple of tries: one song, “Butter,” had a hand-edit export missing its opening chords, caught and re-exported before anything shipped; a few other titles (“Hard Drinking,” “I Need a Lover,” “The Kiss / Da Da Da (Slave to an Angel)”) had drifted in the fresh export and were restored to the established titles.
Week thirteen — the last of the old engine (August 8, 2026)
One show left running on the very first version of the processing engine. Bringing it up to date turned out to correct the setlist as well as the sound.
- Mad Hannans at Cafe Java, Sept 9, 1999 was re-run through the current engine, replacing the original v1 processing — the last show in the archive still on it. This is a difficult audience tape, loud drums and close applause, and the fresh hand-edit pass applied noise reduction across the whole show before any of the automatic work began. Nineteen of its songs needed a target below −20 LUFS to stay truly linear, one as low as −27.1, rather than be squeezed to hit a number.
- A song that was hiding inside another. What the archive listed as a single 5:39 track, “Everything Reminds Me of You,” is in fact two songs played back to back. They’ve been split, and the second one restored under its own name, “Hear Me” — so the show now runs 22 songs instead of 21.
- Two songs got their names back. “My Dear” is the same song the archive lists elsewhere as “Ride On,” and has been retitled to match. Separately, two performances catalogued as “I Need a Lover” — at 19 Broadway in June 1999 and the Sweetwater in Feb 2000 — are also “Hear Me,” and now sit with the song’s other performances instead of on a page of their own.
- Smaller corrections that came along with it: “Plastic Melons” to “Plastic Lemons,” and a stray capital in “I Thought I Was You.”
Week fourteen — a sharper tool for close-mic’d drums (August 8, 2026)
The linear-normalization fix from a few weeks back drew a hard line: if a track can’t hit the archive’s loudness target without risking flattened dynamics, it lands quieter instead. Right for a fade or a fingerpicked verse — but on tapes where the mic sat close to the drums, that same rule was punishing songs whose only loud moments were brief, sparse hits, taking the whole track several dB below target to protect a handful of instants. Two rounds of loudness-matched blind listening — on this show and, independently, a second one at the Sweetwater — found no audible difference from a limiter narrow enough to act only on those instants, three orders of magnitude below the fade-flattening timescale the original ban exists to prevent. That became a new, strictly opt-in engine mode: a millisecond-scale true-peak cap, used only where a track’s loud moments are genuinely sparse, with a hard 6 dB ceiling on how much any single instant can be shaved and the rest of the performance left at its full, untouched gain.
- Mad Hannans at Cafe Java, Sept 9, 1999 — the same show re-split just weeks ago — is the first to ship with the new mode. Fifteen of its twenty-two tracks now reach the full −20 LUFS standard this way instead of landing quieter; “Rocky Road to Dublin / Star of County Down” needed more recovery than the 6 dB ceiling allows and lands at −21.6 instead — inaudibly quieter, never over-shaved. “Truck” and “Anna May” are untouched by the new mode; both were already handled correctly by the applause-only limiter from the earlier fix, which still takes precedence whenever it applies. Every track’s treatment — capped, applause-limited, or plain linear — is recorded in its own audit trail, visible per track on the show page.
- Two titles picked up small spelling corrections back to the archive’s established convention: “ABC (alt version)” and “The Kiss / Da Da Da (Slave to an Angel).”
- Jerry Hannan solo at 19 Broadway, Jan 15, 2001 is the second show to ship with the new mode, replacing its v7 pass. Twenty-one of its thirty-one songs now reach the full target this way; “Woman” stays on the applause-only limiter, unaffected by the new mode. The fresh hand-edit export had drifted track names for four songs — restored to the archive’s established titles, “Highway Patrolman,” “Everything Reminds Me of You,” “The Barney Stone Blues,” and “Woman,” rather than overwritten.
- Mad Hannans at Sweetwater, Jan 6, 2001 is the third show to ship with the new mode, replacing its v2 corrective-EQ pass — the same restoration EQ, cutting low-mid mud and lifting presence and air on this muddled tape, carries over unchanged. Two songs, “Hear Me” and “Rocky Road to Dublin,” now reach the full target this way; “Far Away Eyes” needed more recovery than the 6 dB ceiling allows and lands at −23 instead of −20 — inaudibly quieter, never over-shaved. The fresh export had drifted two more track names, restored to the archive’s established titles: “Hear Me” and “The Kiss / Da Da Da (Slave to an Angel).”
- Mad Hannans at New George’s, Oct 13, 1999 is the fourth show to ship with the new mode, replacing its v2 pass. Six of its fourteen songs now reach the full target this way, most notably “Anna May” and “Butter,” both of which would otherwise have landed several dB short. The fresh export had drifted one more track name, restored to the archive’s established title: “Hear Me.”
- Jerry & Sean Hannan at 19 Broadway, Dec 1999 is the fifth show to ship with the new mode, replacing its v2 pass. Eight of its thirty songs are handled by the applause-only limiter and eleven more qualify for the transient cap, together bringing twenty-one of thirty songs up to the full target instead of landing quieter. Four capped tracks needed a listen before shipping and checked out as legitimate music, not defects. Four track titles were restored to the archive’s established spelling after the fresh export drifted: “Woman,” “Seven Drunken Nights,” and “The Kiss / Da Da Da (Slave to an Angel)” among them.
- Mad Hannans at 4th Street Tavern, May 1, 1999 is the sixth show to ship with the new mode, replacing its v2 pass. Two songs, “Soundcheck” and “Rocky Road to Dublin,” qualify for the transient cap and now reach the full target, both clearing automatically with no listen required. Four track titles were restored to the archive’s established spelling after the fresh export drifted, including “Hear Me” and “The Kiss / Da Da Da (Slave to an Angel).”
- Jerry Hannan solo at 19 Broadway, March 29, 1999 is the seventh show to ship with the new mode, replacing its v2 pass — the last show in the archive on the original engine. Its Drive folder held a second problem along the way: two full upload passes of the noise-reduced export, months apart, had left 34 stale duplicate files from the first pass sitting alongside the real one. Once identified by timestamp and cleared, five of its thirty-four songs came up on the applause-only limiter and sixteen more qualified for the transient cap, bringing twenty-nine of thirty-four up to the full target. Three capped tracks and two suspected clicks all checked out as legitimate music by ear before shipping. One title was restored to the archive’s established spelling, “Hear Me,” and one track was freshly retitled “Limerick Medley” after a listen, since neither the old nor new name matched anything else in the archive to cross-reference against.
- With the archive-wide rollout to the new engine complete, Mad Hannans at Sweetwater, May 18, 1999 became the first show opted back in voluntarily for the transient-cap mode alone, replacing its v5 pass — the corrective EQ from that tape’s earlier restoration, brightening a muddy room sound, carries over unchanged. Fifteen of its twenty-two songs now reach the full target this way. One, “Rocky Road to Dublin,” needed a listen before shipping and checked out as audience handclaps, not a defect; a second track, “Smoke in Heaven,” had similar handclap moments flagged during diagnose, also confirmed legitimate. Three titles were restored to the archive’s established spelling after the fresh export drifted: “Hear Me,” “The Good Life,” and “The Kiss / Da Da Da (Slave to an Angel).”
- Jerry Hannan’s soundboard show at 19 Broadway, Jan 8, 2001 is the second voluntary opt-in, replacing its v7 pass. Sixteen of its thirty songs qualify for the transient cap and two more sit on the applause-only limiter, together reaching the full target instead of landing quieter. Nine capped tracks needed a listen before shipping and all checked out as legitimate music. “The Wind” needed more recovery than the 6 dB ceiling allows; opted into partial capping after confirming the loud moments are music, it lands at −22.65 instead of −20 — inaudibly quieter, never over-shaved. One title was restored to the archive’s established spelling: “Everything” to “Everything Reminds Me of You.”
- “The Wind” got a follow-up, one-off exception to the standard 6 dB cap after a loudness-matched listening test at 6/8/10 dB of cut depth on its loudest moment (a vocal peak) found no audible difference — it now reaches the full −20 target the same way its neighbors do. This is a per-track decision, not a policy change: the engine now records the exact ceiling used on every capped track and flags any exception explicitly, so this stays auditable rather than resting on memory.
- Sean Hannan solo at 19 Broadway, Nov 29, 1999 is the third voluntary opt-in, replacing its v6 pass. Eleven of its twenty-one songs qualify for the transient cap; two needed a listen before shipping and checked out as legitimate music. “The Kiss - Da Da Da (Slave to an Angel)” needed more recovery than the 6 dB ceiling allows and was opted into partial capping. One title was restored to the archive’s established spelling and a stray trailing space trimmed from another.
- Jerry Hannan solo at Cafe Java, April 8, 1999 is the fourth voluntary opt-in, replacing its v4 pass. Five of its twenty-three songs qualify for the transient cap, reaching the full target instead of landing quieter. Two capped tracks — “Anna May” and “Everything Reminds Me of You” — needed a listen before shipping and checked out as legitimate, louder musical passages. One title was corrected: the fresh export’s filename read “Angel of Montgomery,” a mechanical typo not used anywhere else in the archive; kept the established spelling, “Angel from Montgomery.”
- Jerry Hannan solo at 19 Broadway, Feb 1, 1999 is the fifth voluntary opt-in, replacing its v5 pass — the whole-show noise reduction from that earlier pass carries over unchanged. Five of its twenty-one songs qualify for the transient cap and a sixth, “The Grey Funnel Line,” stays on the applause-only limiter. “Hello in There” needed 6.2 dB of recovery, just over the standard 6 dB ceiling; given a one-track exception after review, it now reaches −20.9 instead of the −26.2 it would have landed at otherwise. Two flagged moments were checked and aren’t defects: a mid-song high-crest window in “Daddy” is mic handling noise, and a suspected click in “Everything Reminds Me of You” is an audience clap. “Blind Man” has a real gap in the tape that can’t be repaired by processing — flagged on the show page rather than papered over. One title was corrected back to the archive’s established spelling, “Angel from Montgomery,” the same typo already caught on another show.
- Jerry Hannan solo at 19 Broadway, June 7, 1999 is the sixth voluntary opt-in — and the first to reprocess only the tracks that actually needed it, rather than the whole show. Twelve of its twenty-nine songs qualified for the transient cap; the other seventeen were already sitting at their loudness target under the old v5 pass and gained nothing from a fresh render under v8, so they were left untouched entirely — same audio, same file, same everything, still on their original version. Five capped tracks needed a listen before shipping and all checked out as legitimate music. Two more, “Hello in There” and “Hollywood (alt verse),” declined the cap automatically for exceeding the calibrated review band; after listening, both were given an explicit override — “Hello in There” now reaches the full target, “Hollywood” recovers to within a third of a decibel of it, capped out at the 6 dB ceiling. One title was corrected: the fresh export read “Come On All You Young Maidens” for a song the archive (and this show’s own description) already knows as “A Bunch of Thyme” — that track was already at target from the original pass, so only its filename needed fixing, not its audio.
Week fifteen — a bug that had been flattening the music (August 11–13, 2026)
The whole point of the archive’s loudness rule is that a performance gets raised by one constant amount — a quiet verse and a loud chorus keep the same distance between them, exactly as played. On August 11 a long-standing bug surfaced in the older versions of the processing tool: when the gain needed to reach the target would have pushed a peak too high, the tool quietly switched to adjusting the volume moment by moment instead — the very thing the archive’s rules forbid — and said nothing in its logs. Tracks processed that way had been sitting in the archive sounding subtly flatter than the tape.
- Nothing was audibly broken, which is why it lasted. The effect is a squashing of light and shade rather than an obvious fault, and the tool reported success either way. It was caught by checking what the engine actually did against what it claimed to do, not by listening for something wrong.
- Every show still carrying tracks from the affected versions was re-run, from the earliest Mad Hannans set at Marin Brewing Co. in April 1998 through the Cafe Java and 19 Broadway runs of 1999. The current engine cannot fail this way even in principle: it measures with one tool and applies the gain with another that has no moment-by-moment mode to fall back to.
- A follow-up pass brought the stragglers up. Between the re-runs and this second pass, nineteen shows were touched over the three days. Re-running them left a tail of tracks sitting a decibel or two below the archive’s standard — safe, but quieter than their neighbours for no musical reason. Each was reviewed individually and, where the loud moments turned out to be sparse instants rather than sustained playing, given the narrow transient cap so it could reach the full target. A handful needed the standard 6 dB shaving limit lifted slightly; every one of those is an explicit, recorded decision rather than an automatic one.
- Where the archive landed: of its 680 songs, 598 now sit within half a decibel of the −20 LUFS standard. The remainder are quieter on purpose — tapes whose peaks genuinely can’t be raised further without shaving the music, which stays the line that isn’t crossed.
Week sixteen — one player behind every play button (August 13–18, 2026)
The site had quietly grown two separate pieces of playback machinery — one driving the show pages, another the playlists — which meant every improvement had to be made twice and every bug fixed twice. They are now a single shared engine, and the older one has been deleted rather than left lying around.
- Show pages, song pages and playlists all run the same player now. Nothing about it should look different; the point is that it now behaves identically everywhere, and future work only has to be done once.
- Seeking on a phone actually works. Dragging to a spot in a song meant landing a fingertip inside a 3 pixel-high strip. The target is now 24 pixels tall — the accessibility standard’s minimum — while the thin line you see stays exactly as thin.
- The site’s visitor counter had never been running. The page’s own security policy was blocking the script that collects it, silently, for as long as it had been switched on. Worth recording because the vendor’s own documented fix would not have worked either.
Week seventeen — a louder way to listen (August 16–19, 2026)
The −20 LUFS standard the archive is mastered to was chosen for comfortable listening on real speakers with the dynamics left intact. On a phone speaker, or in a car, it is simply too quiet — and most people never go looking for a volume setting. So every one of the 680 songs was given a second, louder rendering at −14 LUFS, and that is now what plays by default when you press play.
- The archive itself is untouched. The −20 masters are still the real thing, still what the technical data on each show page describes, and still what you get when you download. The louder version is a copy made from those masters, never from the original tapes again, so it cannot drift away from them — same hand-drawn fades, same noise reduction, same edits.
- Every page with a player says which version you are hearing, and a switch on the page moves between them. That disclosure is the condition on which the louder version became the default: it is a real trade, and it would be wrong to make it quietly.
- It was tested by ear before it shipped, and the test was checked too. The eight tracks the louder rendering changed most, plus two decoys where both sides were secretly identical, were compared blind at matched volume — no difference was audible on any of them. Then one of those tracks was deliberately over-compressed to two greater degrees to prove the test could detect the thing it was looking for: both of those were heard. The honest conclusion is a bounded one — inaudible at the level actually shipped, audible somewhere above it — which is why the setting isn’t being pushed any further.
- The louder version can be downloaded too, as of August 19. It is never the preselected choice, though, and the site deliberately forgets which one you picked last time: the two are different formats, not just different volumes — lossless FLAC against a 320 kbps MP3 — and a remembered preference would quietly hand out lossy files for the rest of a visit. A louder download also renames its own folder and carries a note inside, so an unpacked copy can never be mistaken for the master.
Week eighteen — saying less, and saying it where it matters (August 18–19, 2026)
A pass over the pages themselves, cutting standing text that never changed from show to show. Copy that says the same thing on all thirty pages stops being read; worse, it can crowd out the details that genuinely differ.
- One fact had been stated four separate times on a single show page — that lossless downloads are password protected. It now appears once, plus on the download button itself, where it is actually needed.
- A status badge reading “done” was removed from every show page. It appeared on all of them, which made it decoration wearing the costume of data.
- The full-show cards now name both files and both sizes — the lossless download and the streaming version, each with its own figure. There had been a single size on the card with two different files behind it, which invited reading one number as though it described both.
Week nineteen — retiring the pop-up, and one row of controls (August 20, 2026)
More consolidation on the playback side, following straight on from the shared player engine a couple of weeks back.
- The separate pop-up player is gone. It was always a desktop-only thing — a phone opens a tab, not a pop-up, and a backgrounded tab stops playing audio anyway — so it was quietly serving fewer and fewer visitors. The bottom mini-player bar now does that job everywhere, on every kind of screen. An old link to the pop-up still lands you on the Playlist page with the right songs queued up.
- Two experiments were tried and taken back out, honestly. For about a day, the currently-playing row on a show page stuck to the top of the screen as you scrolled, and the Playlist page did the same with its now-playing card. Both worked, and both were removed once it was clear that having a song pinned in two places at once — the sticky row and the bar — read as a bug rather than a feature.
- Show and song pages had grown two separate rows of controls — one for streaming quality, one for select-and-download — in two different visual styles. They’re now a single row of matching buttons. Along the way, 23 show descriptions lost a paragraph of processing boilerplate that had been repeating itself on every one of them, continuing the trim from a couple of weeks back.
- The Playlist builder’s filter list got shorter: the Songwriter filter and four thinly-used genre tags (folk, country, blues, rock — 94 tracks between them across the whole archive) are off the visible list. Nothing was deleted from the catalog or from Search — the “Traditional & Irish” preset still works exactly as before.
Week twenty — easier to find your way around (August 21, 2026)
A run of smaller usability fixes, most of them things Rene ran into using the site himself.
- “Select all” finally toggles. Press it once and every track on the page is selected and the button now reads “Unselect all”; press it again and the selection clears. Before this, a second press simply did nothing.
- A song’s own page now lists its past performances as proper track rows — the same compact look as a show page — instead of a taller, differently-styled player card for each one. On a phone that’s roughly 15 performances visible on screen at once, where it used to be about 6.
- The Search page can now be browsed by filter alone — pick an artist or a venue and see results without typing a word first. The Songs page also gained the “Sean & Jerry” switch its own legend had been showing all along, which the artist filter had quietly left out.
- The password-protected download window now names the file size next to each version you can choose — the lossless archive file and the louder MP3 — instead of one size sitting on the button that used to describe two differently-sized files.
- Fixed a bug going back to when song-picking first launched in July: clearing your selection, or removing every track by hand, left the “N songs selected” bar sitting on screen anyway. A leftover style rule had been overriding the one meant to hide it.
Week twenty-one — a link for one song (August 22, 2026)
Sharing used to mean sending someone a whole show and telling them which track to skip to. Every one of the archive’s 680 curated tracks now has its own short link that opens straight to that one song, already queued and ready to play.
- Getting that link to actually work outside the site took a couple of tries. The link itself first needed its exact address settled, closing a needless extra hop that had passed every check while quietly doing nothing on the live site. Then, trying the whole point of the feature — pasting a link into a Facebook reply — Facebook refused to treat it as a real link at all, for two separate reasons hiding in the page itself: a leftover tag was pointing Facebook at the wrong address, and a “don’t index this” tag meant only for search engines turned out to also tell Facebook’s own crawler to refuse to read the page.
- The share pop-up used to hand over a block of text — song title, artist, venue, date — glued to the link. It now copies just the link; a rebuilt preview card does the describing instead, the way a shared link actually renders in Facebook, iMessage, and the rest.
- Tapping anywhere on a track row now plays it, not only the small play button — matching what every phone music app already trains a thumb to expect. The song title, which used to look tappable but wasn’t, is the biggest part of that fix.
- Fixed the browser-tab icon and the image your phone’s lock screen shows while a track plays, both of which still read “The Hannan Recordings” long after the site itself became “The Hannan Tapes” — and finished retiring the old name from the last couple of places it was still hiding.
- Caught and fixed a two-hour bug where an earlier edit left a stray scrap of code visible at the very top of every single page on the site.
Week twenty-two — one button instead of four (August 23–24, 2026)
Every track row had been carrying four separate small controls — download, share, add to playlist, and a link back to the show — crowding the row and drifting out of alignment with each other. They’re now one “…” menu: a sheet that rises from the bottom on a phone, a small pop-up next to your cursor on a computer. The mini-player bar picked up the same menu, folding in its own separate share button alongside it.
- Two bugs only showed up on an actual iPhone, after every automated check had passed. The “…” button itself was invisible — present, correctly placed, and tappable, just drawing nothing, a quirk of how Safari sizes an icon with no explicit dimensions. And the Download row inside the open menu was shorter than every row beside it, a left-over style collision from an older button. Both are fixed, and both are now specifically checked for before anything ships.
- A fresh outside review turned up four real, if quiet, bugs. “Select all” had gone dead on show and song pages — it was reading off a button the row-menu clean-up had just deleted. Song-page performance rows could show a blank play icon with nothing in it. The Search page mislabeled its own results as “680 songs” when that number actually counts individual performances of 136 distinct songs — and it had been calculating a match-quality score for each result without ever using that score to sort them, so the best match wasn’t necessarily shown first.
- “Play random tape,” on the homepage, wasn’t actually picking a random tape. It picked a random track out of all 680, so it would land you mid-set on a guitar solo or a between-song aside, and it quietly favored whichever shows happened to have the most songs. It now genuinely picks a random show and starts it from the first track, the way it always should have.
- A batch of accessibility fixes came out of the same review: the password download window now properly traps keyboard focus while it’s open and hands focus back to whatever you clicked when it closes; pages that update their own result counts announce the change to a screen reader; and several controls that were missing real labels got them.
- One more bug worth naming, because it took two tries to actually fix. The round handle on a track’s progress bar was visibly sliced in half by the row’s own rounded corner — on the very first and last row of any track list, most noticeable on song pages. The first attempt only moved the handle sideways; the clipping was happening vertically too, and it took a second pass to actually clear it.
- Shared songs were still showing the retired name. The preview card that appears when you paste a link into a message had been redrawn two days earlier, and the corrected version was sitting on the server the whole time — but a link preview is fetched separately from the page and then kept, and redrawing a picture gives a phone that already has the old one no reason to go and look again. The card’s address now changes whenever the picture does, so a new one actually arrives. The image your lock screen shows during playback works the same way and got the same treatment.
Week twenty-three — the first new tape (August 30, 2026)
Every recording in this archive until now came from a five-year run of Marin County clubs between 1998 and 2001. This one was made this week: Jerry, solo, two sets at the Inn at Occidental in Sonoma County, twenty-one songs.
- Five of these songs appear nowhere else in the archive — “Society,” “Girl Shoes,” “Never Be the Same,” “Brother Know,” and “Love is All.” “Society” is the one Eddie Vedder recorded for the Into the Wild soundtrack, and until now it had never turned up on a tape here.
- “Never Know Love” had been caught on tape exactly once before — in Sean’s 19 Broadway set from January 2000, which makes this the only recording of Jerry playing it. And “Perfect Autumn Day,” taped four times across 1999 — once as the finished song and three times as an early version — hadn’t been recorded since.
- The evening was recorded as two separate sets, in two separate projects. They’re published as one show in one running order, the way it was played.
- Applause turned out to be the loudest thing on four of the songs. At the end of “The Parting Glass” the clapping peaks about seventeen decibels above the loudest note Jerry actually plays. Left alone, that one burst would have held the entire song down to a noticeably quieter level than everything around it — so the applause alone is eased back, and the music comes up to where it belongs.
The work continues — more shows, better audio, and small fixes are always in progress.