现已上线！2025 年 Telegram 研究 — 年度关键洞察 
1 641
订阅者
+124 小时
+27 天
+1630 天
数据加载中...
相似频道
标签云
进出提及
---
---
---
---
---
---
吸引订阅者
六月 '26
六月 '26
+9
在0个频道中
五月 '26
+27
在0个频道中
Get PRO
四月 '26
+36
在0个频道中
Get PRO
三月 '26
+20
在1个频道中
Get PRO
二月 '26
+38
在0个频道中
Get PRO
一月 '26
+23
在0个频道中
Get PRO
十二月 '25
+16
在0个频道中
Get PRO
十一月 '25
+25
在0个频道中
Get PRO
十月 '25
+26
在1个频道中
Get PRO
九月 '25
+45
在0个频道中
Get PRO
八月 '25
+18
在0个频道中
Get PRO
七月 '25
+18
在0个频道中
Get PRO
六月 '25
+28
在1个频道中
Get PRO
五月 '25
+34
在0个频道中
Get PRO
四月 '25
+34
在1个频道中
Get PRO
三月 '25
+40
在1个频道中
Get PRO
二月 '25
+36
在0个频道中
Get PRO
一月 '25
+20
在1个频道中
Get PRO
十二月 '24
+56
在2个频道中
Get PRO
十一月 '24
+37
在1个频道中
Get PRO
十月 '24
+137
在3个频道中
Get PRO
九月 '24
+47
在2个频道中
Get PRO
八月 '24
+51
在1个频道中
Get PRO
七月 '24
+62
在2个频道中
Get PRO
六月 '24
+34
在1个频道中
Get PRO
五月 '24
+41
在1个频道中
Get PRO
四月 '24
+33
在1个频道中
Get PRO
三月 '24
+55
在2个频道中
Get PRO
二月 '24
+56
在2个频道中
Get PRO
一月 '24
+77
在1个频道中
Get PRO
十二月 '23
+96
在3个频道中
Get PRO
十一月 '23
+47
在2个频道中
Get PRO
十月 '23
+37
在1个频道中
Get PRO
九月 '23
+23
在0个频道中
Get PRO
八月 '23
+24
在0个频道中
Get PRO
七月 '23
+34
在0个频道中
Get PRO
六月 '23
+20
在0个频道中
Get PRO
五月 '23
+79
在0个频道中
Get PRO
四月 '23
+51
在0个频道中
Get PRO
三月 '23
+15
在0个频道中
Get PRO
二月 '23
+6
在0个频道中
Get PRO
一月 '23
+11
在0个频道中
Get PRO
十二月 '22
+12
在0个频道中
Get PRO
十一月 '22
+22
在0个频道中
Get PRO
十月 '22
+14
在0个频道中
Get PRO
九月 '22
+25
在0个频道中
Get PRO
八月 '22
+53
在0个频道中
Get PRO
七月 '22
+9
在0个频道中
Get PRO
六月 '22
+16
在0个频道中
Get PRO
五月 '22
+14
在0个频道中
Get PRO
四月 '22
+26
在0个频道中
Get PRO
三月 '22
+9
在0个频道中
Get PRO
二月 '22
+8
在0个频道中
Get PRO
一月 '22
+10
在0个频道中
Get PRO
十二月 '21
+16
在0个频道中
Get PRO
十一月 '21
+11
在0个频道中
Get PRO
十月 '21
+16
在0个频道中
Get PRO
九月 '21
+3
在0个频道中
Get PRO
八月 '21
+24
在0个频道中
Get PRO
七月 '21
+13
在0个频道中
Get PRO
六月 '21
+16
在0个频道中
Get PRO
五月 '21
+220
在0个频道中
| 日期 | 订阅者增长 | 提及 | 频道 | |
| 11 六月 | 0 | |||
| 10 六月 | +1 | |||
| 09 六月 | +1 | |||
| 08 六月 | 0 | |||
| 07 六月 | 0 | |||
| 06 六月 | +1 | |||
| 05 六月 | +1 | |||
| 04 六月 | 0 | |||
| 03 六月 | 0 | |||
| 02 六月 | +3 | |||
| 01 六月 | +2 |
频道帖子
Recent SynSIG seminars finally uploaded
https://www.youtube.com/@isca-synsig
for example
https://www.youtube.com/watch?v=M8n9I9eGyTM
| 2 | Deepmind paper
SURF: Separation via Unsupervised Remixing Flow
https://google.github.io/df-conformer/surf/
https://arxiv.org/abs/2606.04921
The goal of single-channel source separation is to reconstruct K sources given their mixture. In supervised settings where vast amounts of clean source data are available, this challenging, illposed problem has been addressed successfully by generative diffusion and flow-based prior models. However, access to such clean source samples is often limited. To bridge this gap, we present Separation via Unsupervised Remixing Flow (SURF), an unsupervised flow matching approach for source separation that learns directly from observed mixtures. This method relies on a novel combination of state-of-the-art supervised flow matching and regression-based selfsupervised techniques. At a high level, starting from a teacher model, we utilize a “remixing” step to bootstrap the learning of a student flow model from the teacher’s estimates. We provide insights into the objectives optimized by this approach and draw a novel connection to the Wake-Sleep algorithm. Empirical evaluations on image and audio benchmarks demonstrate that SURF establishes a new state-of-the-art, significantly outperforming existing unsupervised methods. | 453 |
| 3 | Everyone looks for translation these days. Good paper covering the task complexity (naturalness, prosody, content). Omni systems still unrealistic
https://arxiv.org/abs/2606.03241
Benchmarking Speech-to-Speech Translation Models
Alkis Koudounas, Hayato Futami, Quentin Jodelet, Osamu Take, Shinji Watanabe, Emiru Tsunoo
Speech-to-speech translation (S2ST) has advanced rapidly, but offline evaluation lacks a unified protocol: studies report non-overlapping metric subsets, preventing direct comparisons. We introduce COMPASS, a unified and reproducible benchmarking framework integrating 46 metrics across eight dimensions, and deploy it on 1,248 model-language configurations from FLEURS and CVSS, spanning cascaded and end-to-end architectures over ten language pairs. Architectures exhibit complementary strengths: best-vs-worst gaps exceed 30\% on naturalness and speaker preservation but remain within a few points on translation quality, so single-metric rankings systematically misrepresent system quality. Correlation filtering reduces 46 metrics to 10 per direction, with three axes requiring different metrics across XEN and ENX (e.g., TER/UTMOS vs. ChrF++/NISQA-MOS); these subsets preserve rankings (Spearman's ) while cutting evaluation time by . Human validation across dubbing, podcasts, and medical domains shows standalone MOS predictors fail to predict listener preference, while top domain-specific metrics correlate with human judgment (). We release COMPASS as a foundation for domain-aware S2ST evaluation. | 692 |
| 4 | Things go fundamental
https://github.com/xzf-thu/Audio-Interaction/
Large-scale streaming-audio dataset for audio-LLM / audio-agent training. Each row is a stream: a sequence of audio turns sharing one unified schema. ~2.28M unique audio clips are organised into six task subsets.
https://huggingface.co/datasets/zhifeixie/StreamAudio-2M | 704 |
| 5 | https://aslp-lab.github.io/SmartGlasses/ | 1 034 |
| 6 | https://github.com/harrrshall/natscore
Modern neural TTS (CosyVoice2, F5-TTS, MaskGCT, Llasa, XTTS-v2, etc.) generates speech that crosses the threshold where the dominant failure mode is no longer artifacts; it's subtle unnaturalness: prosody glitches, expressive overshoot, speaker-clone drift, breath placement, code-switching mismatches. The existing automatic naturalness scorers were not trained on this kind of failure surface.
UTMOSv2 (the VoiceMOS 2024 winner) was trained on read-speech MOS labels. It saturates at high quality and is documented to produce negative correlations with human judgment on conversational and expressive speech (arXiv 2603.01467).
WhiSQA was designed for telecom and speech-enhancement quality (NISQA training data). It is intentionally not a synthetic-TTS scorer.
DNSMOS, NISQA-TTS, and the rest of the legacy stack predate modern neural TTS and lack the distribution coverage.
SpeechJudge-GRM (released Nov 2025) is excellent, but it is a 7B-parameter LALM. ~$0.001 per score on Modal A100. Unusable inside a TTS training loop or for large-scale offline evaluation.
The data that fixes the distribution gap, SpeechJudge-Data, was released in November 2025: 99K human-labeled TTS preference pairs across CosyVoice2, F5-TTS, MaskGCT, Llasa, and others, in en/zh + code-switching, with both regular and expressive splits. As of writing, no clean public artifact combines this data with a small, deployable, CPU-runnable scorer.
NatScore fills that gap. | 1 029 |
| 7 |  Some recent training results from our system
https://alphacephei.com/nsh/2026/05/24/asr-details.html | 1 467 |
| 8 | Finally a good competition timeline, not two weeks to implement everything
https://saigonaihub.com/OneVoiceAIChallenge
Dare to build the next generation of realtime translation devices powered by Edge AI
Presented by Saigon AI Hub and Qualcomm
🟠 24 May - 24 June 2026: Registration period
🟠 July 2026: Technical specification submission
🟠 August - September 2026: Prototype submission
🟠 October 2026: Field testing
🟠 November 2026: Grand finale at VNG Campus | 908 |
| 9 | If you decompose prosody you can do many nice things
https://arxiv.org/abs/2605.05927
Minimizing Modality Gap from the Input Side: Your Speech LLM Can Be a Prosody-Aware Text LLM
Wenqian Cui, Xiao-Hui Li, Daxin Tan, Qiyong Zheng, Irwin King
Speech large language models (SLMs) are typically built from text large language model (TLM) checkpoints, yet they still suffer from a substantial modality gap. Prior work has mainly attempted to reduce this gap from the output side by making speech generation more text-like, but the gap remains. We argue that the key remaining bottleneck lies on the input side. We propose TextPro-SLM, an SLM that makes spoken input more closely resemble that of a prosody-aware text LLM. TextPro-SLM combines WhisperPro, a unified speech encoder that produces synchronized text tokens and prosody embeddings, with an LLM backbone trained to preserve the semantic capabilities of the original TLM while learning paralinguistic understanding. Experiments show that TextPro-SLM achieves the lowest modality gap among leading SLMs at both 3B and 7B scales, while also delivering strong overall performance on paralinguistic understanding tasks. These gains are achieved with only roughly 1,000 hours of LLM training audio, suggesting that reducing the modality gap from the input side is both effective and data-efficient. | 1 065 |
| 10 | ASR robusness still a large area to improve
https://github.com/xzf-thu/Mega-ASR | 1 432 |
| 11 | Interesting comments by Desh on TML
https://x.com/rdesh26/status/2054246456635150744
for example Game-Time
Game-Time: Evaluating Temporal Dynamics in Spoken Language Models
https://arxiv.org/abs/2509.26388
. | 1 347 |
| 12 | Following on this, NVIDIA implements transformer encoder instead of conformer
https://github.com/NVIDIA-NeMo/NeMo/pull/15661 | 1 530 |
| 13 | This is quite insightful paper
https://arxiv.org/abs/2601.20094
T-Mimi: A Transformer-based Mimi Decoder for Real-Time On-Phone TTS
Haibin Wu, Bach Viet Do, Naveen Suda, Julian Chan, Madhavan C R, Gene-Ping Yang, Yi-Chiao Wu, Naoyuki Kanda, Yossef Adi, Xin Lei, Yue Liu, Florian Metze, Yuzong Liu
Neural audio codecs provide promising acoustic features for speech synthesis, with representative streaming codecs like Mimi providing high-quality acoustic features for real-time Text-to-Speech (TTS) applications. However, Mimi's decoder, which employs a hybrid transformer and convolution architecture, introduces significant latency bottlenecks on edge devices due to the the compute intensive nature of deconvolution layers which are not friendly for mobile-CPUs, such as the most representative framework XNNPACK. This paper introduces T-Mimi, a novel modification of the Mimi codec decoder that replaces its convolutional components with a purely transformer-based decoder, inspired by the TS3-Codec architecture. This change dramatically reduces on-device TTS latency from 42.1ms to just 4.4ms. Furthermore, we conduct quantization aware training and derive a crucial finding: the final two transformer layers and the concluding linear layers of the decoder, which are close to the waveform, are highly sensitive to quantization and must be preserved at full precision to maintain audio quality.
transformers are faster than CNN
Previous paper from the same author
https://arxiv.org/abs/2411.18803 | 1 410 |
| 14 | Apptek recently released callcenter dataset (129 hours role played). Qwen3-ASR-1.7B leads again
https://huggingface.co/datasets/apptek-com/apptek_callcenter_dialogues | 974 |
| 15 |  HF introduced private leaderboard
https://huggingface.co/blog/open-asr-leaderboard-private-data
Qwen is really good for English | 963 |
| 16 | For people interested in Georgian we also recommend to check
https://huggingface.co/NMikka
good work on finetuning major TTS engines (kokoro, qwen, f5, omni) goes there | 1 113 |
| 17 |  We release Georgian models for old Vosk. Not very outstanding but a good start
https://alphacephei.com/vosk/models/vosk-model-small-ka-0.42.zip
https://alphacephei.com/vosk/models/vosk-model-ka-0.42.zip
https://github.com/alphacep/awesome-speech/blob/main/georgian.md | 1 522 |
| 18 | NAR-CTC thing is interesting, fast
https://www.linkedin.com/feed/update/urn:li:activity:7455374206114115585/
The IBM worldwide speech research team is super excited to announce the release of three Granite 4.1 speech models (links in first comment) that explore different tradeoffs in terms of accuracy, throughput, and functionality:
⛰ granite-speech-4.1-2b (codename "Altius") adds punctuation and truecasing capabilities while also improving multilingual ASR accuracy via a novel dual-head graphemic/BPE CTC encoder with frame importance sampling
💪 granite-speech-4.1-2b-plus (codename "Fortius") adds speaker-attributed ASR, word-level time-stamps and improved support for long-form ASR
⚡️ granite-speech-4.1-2b-nar (codename "Citius") uses a new non-autoregressive architecture that performs parallel local edits to the CTC encoder hypothesis resulting in much higher throughput
🥇 Models 1 and 3 claimed first and third place on the #OpenASR leaderboard. This was truly a worldwide team effort and I would like to express my sincere gratitude to all my IBM colleagues who made this possible: Samuel Thomas, Vishal Sunder., PhD, Jeff Kuo, Brian Kingsbury, Avihu Dekel, Zvi Kons, Hagai Aronowitz, Ron Hoory, Sashi Novitasari, Takashi Fukuda, Tohru Nagano, Masayuki Suzuki, Gakuto KURATA, Madison Lee, Luis Lastras and many more (apologies if I missed you!).
🔬 I hope to see many of you next week at #ICASSP2026 in Barcelona to chat about these models and share some funny anecdotes. | 1 157 |
| 19 | The code for Facebook's LST finally now available
https://github.com/facebookresearch/lst
https://t.me/speechtech/2195 | 0 |
| 20 | For a long time AudioSet was big pain to download, finally available on HF
https://huggingface.co/datasets/agkphysics/AudioSet
Overall, even small speech models need to understand non-speech sounds better. More on this later. | 0 |
