From 12 till 19 November, Aman and I, together with people from VU University (Amsterdam), TNO (Amsterdam) and CSIR (South Africa) were in Mali to work with our colleagues from Sahel Eco, as part of the EU-funded VOICES project.
During our last visit in January 2011, we focused on collecting requirements (see the report), our objective this time is the deployment of the first version of the radio platform we are working on.
During this week, we visited our two radio station partners (Radio ORTM Segou in Segou and Radio Moutian in Tominian), and met with farmers and producers groups — some involved in the current system (the Tominian group), some not being involved (the Segou group). Our main objective was to show and train radio people on the voice-based interface to access agriculture-related content to broadcast, validate with them the quality of the audio files our system is generating, the quality of the audio received through the voice platform, and the design of the broadcast interface for them to use the audio content live directly from the voice-platform.
In terms of deployment, we are using two different and independent platforms. One is developed and hosted by Orange in Mali as part of VOICES and is called EMERGINOV. It is an integrated platform based on free and open-sources software elements. The voice platform in EMERGINOV is VoiceGlue, the VoiceXML browser built on top of Asterisk (see a post i wrote long time ago on voice platforms).
The second platform (see the picture on the left) is a completely standalone platform, that can work without power, with just a portable solar panel like shown on the picture. The laptop hosts the complete environment (LAMP), plus a Voxeo Prophecy IVR server for the voice platform. For the telephony hardware, we are using an appliance called Office Route with four GSM ports. The Office Route is a very efficient and robust device that I discovered when talking to Freedomfone people (Thanks Brenda!). I believe this is the first ever standalone VoiceXML browser platform that doesn’t even need power!
We also have a third option, a cloud-based solution hosted partly on Web Foundation site and partly on Evolution, Voxeo VoiceXML cloud hosting, but this is not usable directly in Mali due to the limited unstable international bandwidth, and due to the absence of Malian phone number in Evolution. We will open this platform soon as a demonstration platform.
From my perspective, the results of this trip exceeded my expectations. Not only all the audio quality validation tests were positive, but radio people were very excited by the system. In January, we had only demo, now we showed them a platform then can use right now and they are eager to start.
The focus of the current work and the VOICES project is limited to the broadcast of the agriculture information, but people understand very well the potential power of such platform.
Part of the platform includes the option to record content from a mobile phone, and enable the radio to broadcast it. This is the functionality that attracted most attention, as connecting journalists in field with the Radio is a major problem they are experiencing today. We had also lots of discussions on how to extend the platform to provide more services to listeners, and increase the interactivity between radio and listeners. This is slightly outside the scope of our current work, but something we are eager to explore further in the future. All in one, this was very positive, and looks very promising.
Now, we have to see if this interest becomes a practice, and we will follow closely the activity on the platform in the coming weeks.
The other part of our work, meeting with producers and farmers, was also very instructive. In a big part it confirmed our previous findings. For instance, we met with representatives of producer groups who lives in the suburb of Segou, the second major town of Mali, and who were all literate. They use a lot their phones for their work, but only voice communication. Interestingly, the second most used feature of their phone is the calculator. None of them have ever used the SMS function of their phone. None of them have ever used the phone book of their phone. They all have a paper phone book with them where they note the phone numbers and names. We had also confirmation that lots of families have at least one phone. The coverage is still an issue, and places with good reception are clearly marked! One other interesting point was related to sharing phones. While people are happy to lend their phone to someone to help him, the concept of shared phone does not exist. It is considered has a personal item.
Part of the new findings out of these discussions is the importance of buyer requirements. There is a huge opportunity for people to structure in groups so that they can offer one single point of contact for big buyers. While we had the feeling that trading between big buyers and small-scale farmers is a problem, we hadn’t had so far practical evidences.
For our current work, this is stressing the need for buyers’ services in our system, and the ability/functionality to aggregate small-scale producers.
As always, field trips are very informative and very enlightening. We learnt in a week more information, and collected ideas that we can’t think of from our offices in Toulouse, London or Boston. I’m very excited to see how all these discussions will now transforms in new services at our end, and in usage at radio station and people end.
Aman will detail in another post the discussions with farmers, and we will publish in the coming weeks a dedicated report on this trip.
Ps: I will post online soon a demo of the deployed system that one can test directly.