ICASSP Related Work
Use this to audit novelty and positioning. ICASSP is single-blind, so you cite normally — no
anonymized references, no hiding your own prior work. Reopen the current call only for
dual-submission and prior-publication rules. The four-page limit makes related work a
precision instrument, not a survey.
Positioning checks
- Separate the signal-processing novelty (a new estimator, transform, representation,
objective, architecture, or resource) from an engineering improvement.
- Compare against both the IEEE SPS journal literature and the conference literature; a
reviewer from the relevant technical committee knows both.
- Position against the nearest current-cycle work, not a five-year-old baseline; ICASSP
reviewers work in the subfield and notice a stale comparison immediately.
- Because review is single-blind, cite your own prior work in normal first person and make the
incremental delta explicit rather than obscuring it.
Literature lanes table
| Lane |
Typical venues |
What ICASSP reviewers check |
| SPS journals |
IEEE TSP, TASLP, SPL, TIP, TIFS, TMM, TCI, TSIPN, JSTSP, OJSP |
Whether the closest journal result is acknowledged and the delta stated |
| ICASSP + IEEE conferences |
prior ICASSP, ICIP, ICME, SLT, ASRU |
Whether the nearest conference method is compared or distinguished |
| Speech/audio siblings |
Interspeech, WASPAA, EUSIPCO, DCASE |
Whether a paper mis-attributed to ICASSP actually lives here |
| ML venues |
NeurIPS, ICML, ICLR, AAAI |
Whether generic-ML priors are credited without claiming their generality |
A bibliography that cites only ML venues and no SPS journals tells a committee reviewer the paper
may be rediscovering a known signal-processing result — a recognizable weakness that benchmark
strength does not repair.
The sibling-venue attribution trap
Signal-processing landmarks are scattered across venues, and misattributing one is a credibility
hit. Common confusions to check before citing:
- Papers you think are ICASSP that are actually Interspeech (much of the speech-synthesis
and self-supervised-speech canon) or journal papers (TASLP separation work).
- ICIP for image results and EUSIPCO/WASPAA for European and audio-workshop results.
- Verify each cited venue against dblp (
conf/icassp/, journals/) rather than trusting memory;
see ../../resources/exemplars/library.md for the
guard list.
Positioning vignette
Suppose the paper proposes a low-complexity beamformer for a hearing device. Its nearest neighbors:
a TASLP paper with higher complexity, a prior ICASSP paper with a similar structure but no latency
budget, and an ML paper with a black-box network. The novelty sentence should name all three
contrasts — lower complexity than the journal method, an explicit latency budget the prior ICASSP
work lacked, and interpretability the black-box model lacked — in one or two sentences, because
four pages cannot afford a paragraph per contrast.
Concurrent-work judgment
- Independently concurrent arXiv work: cite neutrally, state the technical difference, and avoid
priority claims a reviewer cannot verify.
- Your own workshop or prior-conference version: cite it in first person and state exactly what
this paper adds, since single-blind review does not require hiding it.
- If a venue's archival status is unclear for dual-submission purposes, declare the overlap rather
than gambling on an organizer's interpretation.
Output format
[Novelty type] estimator / transform / representation / objective / architecture / resource
[Nearest 3 works] <work -> venue (verified) -> technical delta>
[Lane coverage] SPS journals / ICASSP+conf / siblings / ML — gaps?
[Attribution risk] <any venue to re-verify on dblp>
[Novelty sentence] <one ICASSP-ready contribution contrast>
Source: brycewang-stanford/Awesome-Journal-Skills → ICASSP-Skills/skills/icassp-related-work/SKILL.md
1---2name: icassp-related-work3description: Use when positioning an ICASSP submission across the signal-processing literature — IEEE SPS journals (TSP, TASLP, SPL, TIP and siblings), sibling conferences (Interspeech, ICIP, EUSIPCO, WASPAA), and ML venues, citing normally under single-blind review, and sharpening the technical delta against the nearest current-cycle work in four pages.4---567# ICASSP Related Work89Use this to audit novelty and positioning. ICASSP is single-blind, so you **cite normally** — no10anonymized references, no hiding your own prior work. Reopen the current call only for11dual-submission and prior-publication rules. The four-page limit makes related work a12precision instrument, not a survey.1314## Positioning checks1516- Separate the **signal-processing novelty** (a new estimator, transform, representation,17 objective, architecture, or resource) from an engineering improvement.18- Compare against both the **IEEE SPS journal** literature and the **conference** literature; a19 reviewer from the relevant technical committee knows both.20- Position against the **nearest current-cycle work**, not a five-year-old baseline; ICASSP21 reviewers work in the subfield and notice a stale comparison immediately.22- Because review is single-blind, cite your own prior work in normal first person and make the23 incremental delta explicit rather than obscuring it.2425## Literature lanes table2627| Lane | Typical venues | What ICASSP reviewers check |28|---|---|---|29| SPS journals | IEEE TSP, TASLP, SPL, TIP, TIFS, TMM, TCI, TSIPN, JSTSP, OJSP | Whether the closest journal result is acknowledged and the delta stated |30| ICASSP + IEEE conferences | prior ICASSP, ICIP, ICME, SLT, ASRU | Whether the nearest conference method is compared or distinguished |31| Speech/audio siblings | Interspeech, WASPAA, EUSIPCO, DCASE | Whether a paper mis-attributed to ICASSP actually lives here |32| ML venues | NeurIPS, ICML, ICLR, AAAI | Whether generic-ML priors are credited without claiming their generality |3334A bibliography that cites only ML venues and no SPS journals tells a committee reviewer the paper35may be rediscovering a known signal-processing result — a recognizable weakness that benchmark36strength does not repair.3738## The sibling-venue attribution trap3940Signal-processing landmarks are scattered across venues, and misattributing one is a credibility41hit. Common confusions to check before citing:4243- Papers you *think* are ICASSP that are actually **Interspeech** (much of the speech-synthesis44 and self-supervised-speech canon) or **journal** papers (TASLP separation work).45- **ICIP** for image results and **EUSIPCO/WASPAA** for European and audio-workshop results.46- Verify each cited venue against dblp (`conf/icassp/`, `journals/`) rather than trusting memory;47 see [`../../resources/exemplars/library.md`](../../resources/exemplars/library.md) for the48 guard list.4950## Positioning vignette5152Suppose the paper proposes a low-complexity beamformer for a hearing device. Its nearest neighbors:53a TASLP paper with higher complexity, a prior ICASSP paper with a similar structure but no latency54budget, and an ML paper with a black-box network. The novelty sentence should name all three55contrasts — lower complexity than the journal method, an explicit latency budget the prior ICASSP56work lacked, and interpretability the black-box model lacked — in one or two sentences, because57four pages cannot afford a paragraph per contrast.5859## Concurrent-work judgment6061- Independently concurrent arXiv work: cite neutrally, state the technical difference, and avoid62 priority claims a reviewer cannot verify.63- Your own workshop or prior-conference version: cite it in first person and state exactly what64 this paper adds, since single-blind review does not require hiding it.65- If a venue's archival status is unclear for dual-submission purposes, declare the overlap rather66 than gambling on an organizer's interpretation.6768## Output format6970```text71[Novelty type] estimator / transform / representation / objective / architecture / resource72[Nearest 3 works] <work -> venue (verified) -> technical delta>73[Lane coverage] SPS journals / ICASSP+conf / siblings / ML — gaps?74[Attribution risk] <any venue to re-verify on dblp>75[Novelty sentence] <one ICASSP-ready contribution contrast>76```7778---7980**Source:** [`brycewang-stanford/Awesome-Journal-Skills`](https://github.com/brycewang-stanford/Awesome-Journal-Skills) → `ICASSP-Skills/skills/icassp-related-work/SKILL.md`
thedixitjain