ICASSP Workflow
Use this as the project-management skill for an ICASSP submission. Replace every date with the
current official timetable and work backward from the paper deadline. ICASSP is an IEEE
Signal Processing Society conference, rebuilt by a new local committee each year; it has no
standing editor-in-chief, and the cost model is registration, with the proceedings paper
published in IEEE Xplore.
The annual clock (2026-cycle shape, verify each year)
- September — full-paper deadline on the CMS portal (ICASSP 2026: 17 Sep 2025; ICASSP 2027
reported ~16 Sep 2026, the next major gate as of 2026-07-09).
- Autumn/early winter — single-blind technical-committee review.
- December — short rebuttal/author-response window (2026 cycle: reported to ~22 Dec 2025), if
the cycle runs one.
- Late winter — notification; accepted-paper revision window (2026: to ~28 Jan 2026).
- Spring — IEEE Xplore camera-ready (PDF eXpress + IEEE copyright form), registration.
- May — the conference; in-person oral or poster presentation (ICASSP 2026 Barcelona, 4-8 May;
ICASSP 2027 Toronto, 16-21 May).
Milestones
- Venue fit — confirm the contribution is genuinely signal-processing and ICASSP-routed, not a
generic-ML paper or a speech paper better suited to Interspeech (see
icassp-topic-selection).
- Evidence lock — freeze the mechanism, the task-matched metric, the standard baseline, the
condition sweep, and seeds.
- Paper deadline — upload the 4+1 PDF with the author list (single-blind), EDICS, and the
ethical-standards statement through CMS.
- Rebuttal — if it runs, reply early and move the one decision-critical objection.
- Notification and camera-ready — integrate reviewer fixes within the accepted claim, pass PDF
eXpress, sign the copyright form, register.
- Conference — present in person; ship the public artifact.
Backward plan from the September deadline
| Weeks out (heuristic) |
Milestone |
| 8+ |
Mechanism frozen; standard corpus and scorer chosen |
| 6 |
Baseline reproduced on the same corpus/split/scorer |
| 4 |
Condition sweep and ablation complete, seeds logged |
| 3 |
Full draft sitting in the IEEE 4+1 two-column template |
| 2 |
Internal mock review by a subfield-fluent colleague |
| 1 |
PDF eXpress pass, author-block/ethics check, figure legibility |
| 0 |
Upload via CMS; download and re-read the stored PDF |
These offsets are planning heuristics — anchor each to the current official timetable, never to a
past cycle.
Track decision (decide before the deadline)
Standard 4+1 -> TPC review, IEEE Xplore proceedings, in-person present
OJSP-ICASSP 8+1 -> OJSP editorial review, open-access journal, present (not in proceedings),
earlier pre-registration deadline
SPS-journal present -> already-accepted SPS-journal paper, not re-reviewed, not in proceedings
Grand Challenge -> short paper via the challenge, present, may appear in proceedings
Failure modes by stage
- Choosing the corpus or metric late forces a scramble that a subfield reviewer sees through.
- Leaving format conversion to the final week surfaces four-page overflow too late; IEEE
two-column is tighter than a working draft.
- Skipping the mock review forfeits the one chance to hear "that is the wrong metric for this
task" from a friend rather than a reviewer.
- Missing a track's separate earlier deadline (OJSP pre-registration, grand challenge) forecloses
the best-fit path.
Coordination notes
- Assign one named owner for the CMS submission (format, EDICS, metadata) and another for the
reproducibility package; shared ownership is how both slip.
- Archive the exact submitted PDF, since a rebuttal must quote it precisely.
Output format
[Current stage] idea / experiments / writing / submission / rebuttal / accepted / camera-ready
[Next official deadline] <date and source, or unknown>
[Track] standard / OJSP-ICASSP / journal-present / grand-challenge
[Critical path] <three tasks that determine readiness>
[Owner map] <task -> person or role>
Currency note
The 2026-cycle dates and the ICASSP 2027 September gate were checked 2026-07-09 (see
../../resources/official-source-map.md); the 2027 exact deadline is 待核实 until its CFP is
readable directly. Nothing carries from Barcelona to Toronto automatically — re-open the current
call.
1---2name: icassp-workflow3description: Use when planning an ICASSP project timeline — the annual clock from venue fit through the September paper deadline, single-blind review, the short winter rebuttal, spring notification, IEEE Xplore camera-ready, and the May conference — with backward-planning offsets and the choice among standard, OJSP-ICASSP, and grand-challenge tracks.4---56# ICASSP Workflow78Use this as the project-management skill for an ICASSP submission. Replace every date with the9current official timetable and work backward from the paper deadline. ICASSP is an **IEEE10Signal Processing Society conference**, rebuilt by a new local committee each year; it has no11standing editor-in-chief, and the cost model is registration, with the proceedings paper12published in IEEE Xplore.1314## The annual clock (2026-cycle shape, verify each year)1516- **September** — full-paper deadline on the CMS portal (ICASSP 2026: 17 Sep 2025; **ICASSP 202717 reported ~16 Sep 2026, the next major gate as of 2026-07-09**).18- **Autumn/early winter** — single-blind technical-committee review.19- **December** — short rebuttal/author-response window (2026 cycle: reported to ~22 Dec 2025), if20 the cycle runs one.21- **Late winter** — notification; accepted-paper revision window (2026: to ~28 Jan 2026).22- **Spring** — IEEE Xplore camera-ready (PDF eXpress + IEEE copyright form), registration.23- **May** — the conference; in-person oral or poster presentation (ICASSP 2026 Barcelona, 4-8 May;24 ICASSP 2027 Toronto, 16-21 May).2526## Milestones2728- **Venue fit** — confirm the contribution is genuinely signal-processing and ICASSP-routed, not a29 generic-ML paper or a speech paper better suited to Interspeech (see `icassp-topic-selection`).30- **Evidence lock** — freeze the mechanism, the task-matched metric, the standard baseline, the31 condition sweep, and seeds.32- **Paper deadline** — upload the 4+1 PDF with the author list (single-blind), EDICS, and the33 ethical-standards statement through CMS.34- **Rebuttal** — if it runs, reply early and move the one decision-critical objection.35- **Notification and camera-ready** — integrate reviewer fixes within the accepted claim, pass PDF36 eXpress, sign the copyright form, register.37- **Conference** — present in person; ship the public artifact.3839## Backward plan from the September deadline4041| Weeks out (heuristic) | Milestone |42|---|---|43| 8+ | Mechanism frozen; standard corpus and scorer chosen |44| 6 | Baseline reproduced on the same corpus/split/scorer |45| 4 | Condition sweep and ablation complete, seeds logged |46| 3 | Full draft sitting in the IEEE 4+1 two-column template |47| 2 | Internal mock review by a subfield-fluent colleague |48| 1 | PDF eXpress pass, author-block/ethics check, figure legibility |49| 0 | Upload via CMS; download and re-read the stored PDF |5051These offsets are planning heuristics — anchor each to the current official timetable, never to a52past cycle.5354## Track decision (decide before the deadline)5556```text57Standard 4+1 -> TPC review, IEEE Xplore proceedings, in-person present58OJSP-ICASSP 8+1 -> OJSP editorial review, open-access journal, present (not in proceedings),59 earlier pre-registration deadline60SPS-journal present -> already-accepted SPS-journal paper, not re-reviewed, not in proceedings61Grand Challenge -> short paper via the challenge, present, may appear in proceedings62```6364## Failure modes by stage6566- Choosing the corpus or metric late forces a scramble that a subfield reviewer sees through.67- Leaving format conversion to the final week surfaces four-page overflow too late; IEEE68 two-column is tighter than a working draft.69- Skipping the mock review forfeits the one chance to hear "that is the wrong metric for this70 task" from a friend rather than a reviewer.71- Missing a track's separate earlier deadline (OJSP pre-registration, grand challenge) forecloses72 the best-fit path.7374## Coordination notes7576- Assign one named owner for the CMS submission (format, EDICS, metadata) and another for the77 reproducibility package; shared ownership is how both slip.78- Archive the exact submitted PDF, since a rebuttal must quote it precisely.7980## Output format8182```text83[Current stage] idea / experiments / writing / submission / rebuttal / accepted / camera-ready84[Next official deadline] <date and source, or unknown>85[Track] standard / OJSP-ICASSP / journal-present / grand-challenge86[Critical path] <three tasks that determine readiness>87[Owner map] <task -> person or role>88```8990## Currency note9192The 2026-cycle dates and the ICASSP 2027 September gate were checked 2026-07-09 (see93`../../resources/official-source-map.md`); the 2027 exact deadline is 待核实 until its CFP is94readable directly. Nothing carries from Barcelona to Toronto automatically — re-open the current95call.
brycewang-stanford