View in EDS

Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR

Saved in:
Bibliographic Details
Title: Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR
Language: English
Authors: Zilai Wang, Natarajan Balaji Shankar, Kaiyuan Zhang, Zihan Wang, Abeer Alwan
Source: Grantee Submission. 2026Paper presented at the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2026).
Peer Reviewed: Y
Page Count: 5
Publication Date: 2026
Sponsoring Agency: National Science Foundation (NSF)
Institute of Education Sciences (ED)
Contract Number: R305C240046
Document Type: Speeches/Meeting Papers
Reports - Research
Descriptors: Independent Study, Speech Communication, Speech Evaluation, Word Recognition, Children
DOI: 10.1109/icassp55912.2026.11462835
Abstract: Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. Fine-tuning SSL models on child speech induces shifts in the representation space. We hypothesize that delta SSL embeddings, the differences between embeddings from a fine-tuned model and those from its pre-trained counterpart, encode task-specific information that complements fine-tuned features from another SSL model. We evaluate multiple fusion strategies on the MyST children's corpus with different models. Results show that delta embedding fusion with WavLM yields up to a 10% relative WER reduction for HuBERT and a 4.4% reduction for W2V2, compared to fine-tuned embedding fusion. Notably, fusing WavLM with delta W2V2 embeddings achieves a WER of 9.64, setting a new state-of-the-art among SSL models on the MyST corpus. These findings demonstrate the effectiveness of delta embeddings and highlight feature fusion as a promising direction for advancing child ASR. [This paper was published in: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) Proceedings, 2026, pp. 18417-18421.]
Abstractor: As Provided
IES Funded: Yes
Entry Date: 20260101
Accession Number: ED681686
Database: ERIC
FullText Text:
  Availability: 0
Header DbId: eric
DbLabel: ERIC
An: ED681686
AccessLevel: 3
PubType: Conference
PubTypeId: conference
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Zilai+Wang%22">Zilai Wang</searchLink><br /><searchLink fieldCode="AR" term="%22Natarajan+Balaji+Shankar%22">Natarajan Balaji Shankar</searchLink><br /><searchLink fieldCode="AR" term="%22Kaiyuan+Zhang%22">Kaiyuan Zhang</searchLink><br /><searchLink fieldCode="AR" term="%22Zihan+Wang%22">Zihan Wang</searchLink><br /><searchLink fieldCode="AR" term="%22Abeer+Alwan%22">Abeer Alwan</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="SO" term="%22Grantee+Submission%22"><i>Grantee Submission</i></searchLink>. 2026Paper presented at the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2026).
– Name: PeerReviewed
  Label: Peer Reviewed
  Group: SrcInfo
  Data: Y
– Name: Pages
  Label: Page Count
  Group: Src
  Data: 5
– Name: DatePubCY
  Label: Publication Date
  Group: Date
  Data: 2026
– Name: SourceSuprt
  Label: Sponsoring Agency
  Group: SrcSuprt
  Data: National Science Foundation (NSF)<br />Institute of Education Sciences (ED)
– Name: NumberContract
  Label: Contract Number
  Group: NumCntrct
  Data: R305C240046
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Speeches/Meeting Papers<br />Reports - Research
– Name: Subject
  Label: Descriptors
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Independent+Study%22">Independent Study</searchLink><br /><searchLink fieldCode="DE" term="%22Speech+Communication%22">Speech Communication</searchLink><br /><searchLink fieldCode="DE" term="%22Speech+Evaluation%22">Speech Evaluation</searchLink><br /><searchLink fieldCode="DE" term="%22Word+Recognition%22">Word Recognition</searchLink><br /><searchLink fieldCode="DE" term="%22Children%22">Children</searchLink>
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1109/icassp55912.2026.11462835
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. Fine-tuning SSL models on child speech induces shifts in the representation space. We hypothesize that delta SSL embeddings, the differences between embeddings from a fine-tuned model and those from its pre-trained counterpart, encode task-specific information that complements fine-tuned features from another SSL model. We evaluate multiple fusion strategies on the MyST children's corpus with different models. Results show that delta embedding fusion with WavLM yields up to a 10% relative WER reduction for HuBERT and a 4.4% reduction for W2V2, compared to fine-tuned embedding fusion. Notably, fusing WavLM with delta W2V2 embeddings achieves a WER of 9.64, setting a new state-of-the-art among SSL models on the MyST corpus. These findings demonstrate the effectiveness of delta embeddings and highlight feature fusion as a promising direction for advancing child ASR. [This paper was published in: IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) Proceedings, 2026, pp. 18417-18421.]
– Name: AbstractInfo
  Label: Abstractor
  Group: Ab
  Data: As Provided
– Name: CodeSource
  Label: IES Funded
  Group: SrcInfo
  Data: Yes
– Name: DateEntry
  Label: Entry Date
  Group: Date
  Data: 20260101
– Name: AN
  Label: Accession Number
  Group: ID
  Data: ED681686
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=eric&AN=ED681686
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1109/icassp55912.2026.11462835
    Languages:
      – Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 5
    Subjects:
      – SubjectFull: Independent Study
        Type: general
      – SubjectFull: Speech Communication
        Type: general
      – SubjectFull: Speech Evaluation
        Type: general
      – SubjectFull: Word Recognition
        Type: general
      – SubjectFull: Children
        Type: general
    Titles:
      – TitleFull: Mind the Shift: Using Delta SSL Embeddings to Enhance Child ASR
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Zilai Wang
      – PersonEntity:
          Name:
            NameFull: Natarajan Balaji Shankar
      – PersonEntity:
          Name:
            NameFull: Kaiyuan Zhang
      – PersonEntity:
          Name:
            NameFull: Zihan Wang
      – PersonEntity:
          Name:
            NameFull: Abeer Alwan
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2026
          Titles:
            – TitleFull: Grantee Submission
              Type: main
ResultId 1