{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:DBIV7KCGS6HIRJYP7UAUFELKEQ","short_pith_number":"pith:DBIV7KCG","schema_version":"1.0","canonical_sha256":"18515fa846978e88a70ffd0142916a24114527d52ee2b933dd6140933251f4ae","source":{"kind":"arxiv","id":"2108.01704","version":1},"attestation_state":"computed","paper":{"title":"Bifocal Neural ASR: Exploiting Keyword Spotting for Inference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Ariya Rastrow, Grant P. Strimel, Jonathan Macoskey","submitted_at":"2021-08-03T18:58:39Z","abstract_excerpt":"We present Bifocal RNN-T, a new variant of the Recurrent Neural Network Transducer (RNN-T) architecture designed for improved inference time latency on speech recognition tasks. The architecture enables a dynamic pivot for its runtime compute pathway, namely taking advantage of keyword spotting to select which component of the network to execute for a given audio frame. To accomplish this, we leverage a recurrent cell we call the Bifocal LSTM (BFLSTM), which we detail in the paper. The architecture is compatible with other optimization strategies such as quantization, sparsification, and apply"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2108.01704","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.AS","submitted_at":"2021-08-03T18:58:39Z","cross_cats_sorted":["cs.SD"],"title_canon_sha256":"cfeb5190aec76ab0f7d19317ffc546e66add09a7fb432ef2911322481476ed88","abstract_canon_sha256":"1054929adeb1568322de46ed31a5270dfee5b74fb86ca2a53f7f23048de52346"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:03:09.794443Z","signature_b64":"h+qjiBx84zWSVlxkpWsKUdvg4SyrVhWXlfGzn9JKURMKH6DBBW5vBv9a1uPJzp7zdQtabxLUn+4XPm+6M4qACA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"18515fa846978e88a70ffd0142916a24114527d52ee2b933dd6140933251f4ae","last_reissued_at":"2026-07-05T03:03:09.793915Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:03:09.793915Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bifocal Neural ASR: Exploiting Keyword Spotting for Inference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Ariya Rastrow, Grant P. Strimel, Jonathan Macoskey","submitted_at":"2021-08-03T18:58:39Z","abstract_excerpt":"We present Bifocal RNN-T, a new variant of the Recurrent Neural Network Transducer (RNN-T) architecture designed for improved inference time latency on speech recognition tasks. The architecture enables a dynamic pivot for its runtime compute pathway, namely taking advantage of keyword spotting to select which component of the network to execute for a given audio frame. To accomplish this, we leverage a recurrent cell we call the Bifocal LSTM (BFLSTM), which we detail in the paper. The architecture is compatible with other optimization strategies such as quantization, sparsification, and apply"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2108.01704","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2108.01704/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2108.01704","created_at":"2026-07-05T03:03:09.793974+00:00"},{"alias_kind":"arxiv_version","alias_value":"2108.01704v1","created_at":"2026-07-05T03:03:09.793974+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2108.01704","created_at":"2026-07-05T03:03:09.793974+00:00"},{"alias_kind":"pith_short_12","alias_value":"DBIV7KCGS6HI","created_at":"2026-07-05T03:03:09.793974+00:00"},{"alias_kind":"pith_short_16","alias_value":"DBIV7KCGS6HIRJYP","created_at":"2026-07-05T03:03:09.793974+00:00"},{"alias_kind":"pith_short_8","alias_value":"DBIV7KCG","created_at":"2026-07-05T03:03:09.793974+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ","json":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ.json","graph_json":"https://pith.science/api/pith-number/DBIV7KCGS6HIRJYP7UAUFELKEQ/graph.json","events_json":"https://pith.science/api/pith-number/DBIV7KCGS6HIRJYP7UAUFELKEQ/events.json","paper":"https://pith.science/paper/DBIV7KCG"},"agent_actions":{"view_html":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ","download_json":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ.json","view_paper":"https://pith.science/paper/DBIV7KCG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2108.01704&json=true","fetch_graph":"https://pith.science/api/pith-number/DBIV7KCGS6HIRJYP7UAUFELKEQ/graph.json","fetch_events":"https://pith.science/api/pith-number/DBIV7KCGS6HIRJYP7UAUFELKEQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ/action/storage_attestation","attest_author":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ/action/author_attestation","sign_citation":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ/action/citation_signature","submit_replication":"https://pith.science/pith/DBIV7KCGS6HIRJYP7UAUFELKEQ/action/replication_record"}},"created_at":"2026-07-05T03:03:09.793974+00:00","updated_at":"2026-07-05T03:03:09.793974+00:00"}