ychwanegu manylion atalnodi a priflythrennu / add details on restorin…

…g punctuation and capitalization

inference/server/README.md

@@ -71,4 +71,32 @@ ID Start End Transcript
 1 0.619581589958159 5.170041841004185 mae ganddynt ddau o blant mab a merch
 ```
 
-Mae'r gweinydd yn darparu GUI HTML syml iawn hefyd er mwyn defnyddio/cefnogi'r API uchod. Ewch i http:https://localhost:5511/static_html/index.hml
+Mae'r gweinydd yn darparu GUI HTML syml iawn hefyd er mwyn defnyddio/cefnogi'r API uchod. Ewch i http:https://localhost:5511/static_html/index.hml
+
+
+## Atgyweirio priflythrennau ac atalnodi
+
+Mae canlyniadau o'r model adnabod lleferydd cyson mewn llythrennau bach sydd heb unrhyw fath o atalnodi fel marc cwestiwn, atalnod llawn, cyplysnodau ayb fel yn yr enghraifft uchod - "mae ganddynt ddau o blant mab a merch". Mae modd cysylltu'r gweinydd trawsgrifio gyda weinydd atalnodi (o'n project arall at GitHub - https://github.com/techiaith/docker-atalnodi-server).
+
+Gosodwch y weinydd atalnodi a nodwch ei cyfeiriad we (fel http:https://localhost:5555) a nodwch yr URL o fewn ffeil newydd o'r enw `external_api_urls.py` yn y ffolder `worker`. E.e. 
+
+```python
+$ cat worker/external_api_urls.py
+PUNCTUATION_API_URL = "http:https://localhost:5555/restore"
+````
+
+Ac ail-gychwynwch y gweinydd drwy 
+
+```shell
+$ make down
+$ make up
+```
+
+Bydd y canlyniad i'r profi'r API gyda'r ffeil `speech.wav` yn rhoi testun sydd wedi ei briflythrennu a'i hatalnodi:
+
+```
+$ curl localhost:5511/get_srt/?stt_id=.....
+1
+00:00:00,619 --> 00:00:05,170
+Mae ganddynt ddau o blant, mab a merch.
+```

inference/server/README_en.md

@@ -74,3 +74,30 @@ ID Start End Transcript
 
 
 The server provides a very simple HTML GUI in order to use/support the above API. Go to http:https://localhost:5511/static_html/index.hml
+
+## Restore capitalization and punctuation
+
+Results from the speech recognition model are always in lowercase letters and do not contain any type of punctuation marks such question marks, full stops, colons etc. such as in the example transcription result above - "mae ganddynt dau o blant mab a merch". Therefore, it's now possible to connect the transcription server with a punctuation server that you may have installed from our other project on GitHub - see https://github.com/techiaith/docker-atalnodi-server. 
+
+Simply install the punctuation server and enter its web address (such as http:https://localhost:5555/restore) into a new file named `external_api_urls.py` in the `worker` folder. E.g.
+
+```python
+$ cat worker/external_api_urls.py
+PUNCTUATION_API_URL = "http:https://localhost:5555/restore"
+````
+
+Restart your speech recognition server..
+
+```shell
+$ make down
+$ make up
+```
+
+The result of testing the API with the `speech.wav` file will this time give a transcription that is capitalized and punctuated:
+
+```
+$ curl localhost:5511/get_srt/?stt_id=.....
+1
+00:00:00,619 --> 00:00:05,170
+Mae ganddynt ddau o blant, mab a merch.
+```