Nexmo allow us to setup Stream NCCO object to stream a speech sound to the call receiver. Here is some specification of the allowed sound format for stream url:
MP3:
- MPEG Audio Layer 3, version 2
- Constant bit rate
- Bit rate: 16 Kbps (8, 32, 64 also supported)
- Sampling rate: 16.0 KHz
- 1 channel
- Lossy compression
- Stream size: 10.1 KiB (91%)
- Encoded with LAME 3.99.5
WAV:
- 8 or 16-bit Linear PCM
- G.711 A-law/u-law
- Microsoft GSM
But sometimes the WAV files is bigger and make Nexmo takes time to load the sound file before start playing on the call. We can use FFMPEG to convert WAV files to MP3 that suitable. First download FFMPEG executable here. Then run this command:
ffmpeg -i /path/to/source.wav -vn -ar 16000 -ac 1 -ab 64k -f mp3 /path/to/target.mp3
The default bitrate on command above is 64K which passed on -ab parameter.
You can change the bitrate to 32K or 16K. The smaller bitrate you choose,
the smaller it output size, but also it will affect the quality of the mp3
sound output.