|
Work item:
|
P.MLS
|
|
Subject/title:
|
Quality of speech generated with machine learning techniques
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027 (Medium priority)
|
|
Liaison:
|
3GPP, ETSI STQ
|
|
Supporting members:
|
Orange, Alibaba; TU Berlin
|
|
Summary:
|
P.MLS provides subjective methods for the assessment of speech which has been produced with advanced machine-learning techniques, such as large pretrained transformer-based models (foundation models). Such models are used in speech coding, speech enhancement, text-to-speech synthesis, and speech-based chatbots, amongst others. The perceptual effects of this generation principle are identified, and corresponding subjective test methods are defined. Where appropriate, links to methods in existing Recommendations are provided.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-09-18 14:37:07
|
|
Last update:
2025-09-19 09:41:43
|
