Remove limitations section
Browse files
README.md
CHANGED
|
@@ -90,12 +90,6 @@ What this gives you:
|
|
| 90 |
- **Linking speakers across recordings.** Frame-level linking with e-means finds the same person across an archive of more than a thousand speakers, reaching the loss of Lloyd's k-means several times faster.
|
| 91 |
- **Low and constant compute.** The cost per update does not grow with the meeting (real-time factor about 0.09 on a shared CPU host).
|
| 92 |
|
| 93 |
-
## Limitations
|
| 94 |
-
|
| 95 |
-
Evaluated on VoxConverse, AMI (ihm-mix), ICSI, synthetic LibriSpeech conversations and a VoxCeleb1 archive; English only, no far-field evaluation.
|
| 96 |
-
Thresholds were set on half of each corpus. Keep one free CPU core per live session.
|
| 97 |
-
Do not use for identification of people: speaker ids are anonymous and session-local.
|
| 98 |
-
|
| 99 |
## Licenses
|
| 100 |
|
| 101 |
Code: Apache-2.0 (see `LICENSE`, `NOTICE`). Vendored e-means: MIT (Meta). The models it downloads keep their own licenses
|
|
|
|
| 90 |
- **Linking speakers across recordings.** Frame-level linking with e-means finds the same person across an archive of more than a thousand speakers, reaching the loss of Lloyd's k-means several times faster.
|
| 91 |
- **Low and constant compute.** The cost per update does not grow with the meeting (real-time factor about 0.09 on a shared CPU host).
|
| 92 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
## Licenses
|
| 94 |
|
| 95 |
Code: Apache-2.0 (see `LICENSE`, `NOTICE`). Vendored e-means: MIT (Meta). The models it downloads keep their own licenses
|