{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:D4MT2DXWCJLCPWVXJGSKJTYIYU","short_pith_number":"pith:D4MT2DXW","schema_version":"1.0","canonical_sha256":"1f193d0ef6125627dab749a4a4cf08c50db3b7a2d8ece69cfd80b3b428fee750","source":{"kind":"arxiv","id":"2007.03001","version":2},"attestation_state":"computed","paper":{"title":"Massively Multilingual ASR: 50 Languages, 1 Model, 1 Billion Parameters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Anuroop Sriram, Awni Hannun, Gabriel Synnaeve, Paden Tomasello, Ronan Collobert, Vineel Pratap, Vitaliy Liptchinsky","submitted_at":"2020-07-06T18:43:38Z","abstract_excerpt":"We study training a single acoustic model for multiple languages with the aim of improving automatic speech recognition (ASR) performance on low-resource languages, and over-all simplifying deployment of ASR systems that support diverse languages. We perform an extensive benchmark on 51 languages, with varying amount of training data by language(from 100 hours to 1100 hours). We compare three variants of multilingual training from a single joint model without knowing the input language, to using this information, to multiple heads (one per language cluster). We show that multilingual training "},"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":"2007.03001","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.AS","submitted_at":"2020-07-06T18:43:38Z","cross_cats_sorted":["cs.CL","cs.SD"],"title_canon_sha256":"88f3a201b1a65db1a14a689109512478beebc0fab83991058173f37144fc5840","abstract_canon_sha256":"198e49ed0a4efdb4daa5a50da6b742c719083e6cf03fe26ed593658f762d7986"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:17:08.655536Z","signature_b64":"GglbOeCdrFHTKCMMBEWjcng4lxZWufOqfpWbNw8XDPcMFnpfXSNiWWfEfz4B83RVfAvsa9sUinY1sJkX/1jCAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1f193d0ef6125627dab749a4a4cf08c50db3b7a2d8ece69cfd80b3b428fee750","last_reissued_at":"2026-07-05T01:17:08.655044Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:17:08.655044Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Massively Multilingual ASR: 50 Languages, 1 Model, 1 Billion Parameters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Anuroop Sriram, Awni Hannun, Gabriel Synnaeve, Paden Tomasello, Ronan Collobert, Vineel Pratap, Vitaliy Liptchinsky","submitted_at":"2020-07-06T18:43:38Z","abstract_excerpt":"We study training a single acoustic model for multiple languages with the aim of improving automatic speech recognition (ASR) performance on low-resource languages, and over-all simplifying deployment of ASR systems that support diverse languages. We perform an extensive benchmark on 51 languages, with varying amount of training data by language(from 100 hours to 1100 hours). We compare three variants of multilingual training from a single joint model without knowing the input language, to using this information, to multiple heads (one per language cluster). We show that multilingual training "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2007.03001","kind":"arxiv","version":2},"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/2007.03001/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":"2007.03001","created_at":"2026-07-05T01:17:08.655104+00:00"},{"alias_kind":"arxiv_version","alias_value":"2007.03001v2","created_at":"2026-07-05T01:17:08.655104+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2007.03001","created_at":"2026-07-05T01:17:08.655104+00:00"},{"alias_kind":"pith_short_12","alias_value":"D4MT2DXWCJLC","created_at":"2026-07-05T01:17:08.655104+00:00"},{"alias_kind":"pith_short_16","alias_value":"D4MT2DXWCJLCPWVX","created_at":"2026-07-05T01:17:08.655104+00:00"},{"alias_kind":"pith_short_8","alias_value":"D4MT2DXW","created_at":"2026-07-05T01:17:08.655104+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2205.06175","citing_title":"A Generalist Agent","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2109.01652","citing_title":"Finetuned Language Models Are Zero-Shot Learners","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU","json":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU.json","graph_json":"https://pith.science/api/pith-number/D4MT2DXWCJLCPWVXJGSKJTYIYU/graph.json","events_json":"https://pith.science/api/pith-number/D4MT2DXWCJLCPWVXJGSKJTYIYU/events.json","paper":"https://pith.science/paper/D4MT2DXW"},"agent_actions":{"view_html":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU","download_json":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU.json","view_paper":"https://pith.science/paper/D4MT2DXW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2007.03001&json=true","fetch_graph":"https://pith.science/api/pith-number/D4MT2DXWCJLCPWVXJGSKJTYIYU/graph.json","fetch_events":"https://pith.science/api/pith-number/D4MT2DXWCJLCPWVXJGSKJTYIYU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU/action/storage_attestation","attest_author":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU/action/author_attestation","sign_citation":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU/action/citation_signature","submit_replication":"https://pith.science/pith/D4MT2DXWCJLCPWVXJGSKJTYIYU/action/replication_record"}},"created_at":"2026-07-05T01:17:08.655104+00:00","updated_at":"2026-07-05T01:17:08.655104+00:00"}