Crash fix and Additional function swift app

General information for the app: The app records audio clips, and then sends it to a backend server. The server analyzes the audio clip and send the results back to the app.
Kind of development: Customization of existing app
Description of every page/module: We need to fix an existing iOS App. The app is in practice complete and written in Swift. The app records audio clips, and then sends it to a backend server. The server analyzes the audio clip and send the results back to the app. The result is then displayed to the user. The server analyzes the sound clip, and produce how many times and for how long female / male speakers talk on the sound clip.

Backend server is ready and running on Google Cloud. kRemoteURL variable in Audio Recorder class should point to the API endpoint URL that we provide. The server is up and the app is basically completed.

The app crashes after half a minute. We need you to fix the crash so that the app is fully functional.

The code for the app is available here: https://github.com/doberman/speaker-gender-detect--ios.
You will get the kRemoteURL variable to the API if we have come to an agreement.

The result from the server is in JSON format and looks like this:
{"duration": 2186, "url": "tmp/upload_148b0446d7320b58e80c1c748eb5cbd7.wav", "db": "/root/.voiceid/gmm_db", "selections": [{"speakers": {}, "speakerLabel": "S1", "gender": "F", "speaker": "unknown", "startTime": 15.54, "endTime": 21.91}, {"speakers": {}, "speakerLabel": "S0", "gender": "M", "speaker": "unknown", "startTime": 0.0, "endTime": 15.49}]}
Description of requirements/features: Additional functions:

The user will have the option to add a prerecorded audio clip. The app sends the audio clip to the server and gets a resultat. The user can either choose to send the full audio clip to the server or splitt the audio clip and send it in parts. We want the app to be able to split the audio clip in two seconds parts. The two seconds clips are sent separately. When the entire clip has been sent to the server, the app analyze the results it has received from the server and shows the user the total time on how long women / men spoke.

Case senario:
The user uploads a 12-minute audio file. The user chooses to split the audio file in two sekond parts. The user presses a button. The user sees the total time for how long the men talked and for how long the women talked.
Extra notes: